1

Full title

The role of L1 phonology in L2 morphological production:

L2 English past tense production by L1 Spanish, Mandarin, and Japanese Speakers

Running head

L1 phonology effects on L2 morphology

2

Abstract

This study addresses the issue of L2 acquisition in adult learners by considering L1

phonological influence as part of the explanation for the ubiquitous problem of L2 functional

morphology production. We compare the predictions of Representational Deficit Approaches

and Full Access Approaches in the acquisition of English simple past tense by looking at

morphology production by L1 speakers of Spanish, Mandarin, and Japanese. Common to

these L1s are similar phonological restrictions on complex coda formation, typically found in

simple past forms. Differences arise in L1 syntax; while Spanish and Japanese have syntactic

past, Mandarin does not. Results show that having syntactic past in the L1 is not the only or

main factor in target-like performance. Results also show that in some tasks, native speakers

of Mandarin perform in a native-like way or like the other L2 groups, which questions the

RDA position that adults are unable to acquire new syntactic features.

977 characters with spaces

3

1. Introduction

In the past two decades or so, generative second language (L2) acquisition studies

have addressed the perennial problem of variable L2 functional morphology production as

part of a more general, larger query, namely whether it is possible for adult L2 learners

(L2ers) to acquire new target-like syntactic representation (see e.g., Prévost & White 2000

Franceschina 2002, 2005; Hawkins & Liszka 2003; McCarthy 2008; Slabakova 2008;

Lardiere 2009). This means that variability in target L2 morphological production is

understood by some as an indication of target deviancy in related syntactic representation

(but see Rothman 2007, for discussion). Most recently, this issue has taken the form of

examining the status of a number of uninterpretable parameterized syntactic features in L2

grammars (e.g., the syntactic features for number, tense, wh-movement). Among such

features, the feature for past tense ([upast]) has received considerable attention particularly in

the case of native Mandarin Chinese learners (henceforth, Mandarin) of L2 English.

Different, sometimes mutually exclusive, proposals have been offered to account for the

observed fact that suppliance of English past tense morphology proves difficult for this group

(e.g., Hawkins & Chan 1997 for morphosyntactic deficits, Lardiere 1998 for syntax-

morphology mapping difficulties, Davidson 2005, 2006a, 2006b for L1 phonotactic

constraints, Goad, White & Steele 2003; Goad & White 2006 for prosodic constraints, Solt et

al. 2004 for perception, Hopp 2009; Martohardjono, Valian & Klein 2011; Bonner 2013 for

input factors and processing pressure). Indeed, it is intriguing that past tense morphology is

so challenging for L2 speakers given that it is highly frequent during speech and in written

texts, involves overt morphology (except in a few cases, e.g., ‘cut’-‘cut’), and is explicitly

taught in ESL classrooms from the elementary levels.

Assuming L1 transfer, exploring other explanations seems warranted. Mandarin not

only lacks a syntactic feature for past tense, but also has phonological restrictions on complex

4

coda formation that are incompatible with many English past tense allomorphic forms. The

possibility of L1 phonological transfer accounting better for the patterns of past tense

omission by Mandarin speakers is a reasonable one to entertain.

Herein, we explore the possibility that the trend of variable L2 English past

morpheme production by Mandarin speakers at high levels of L2 proficiency is best

explained by L1 phonological influence as opposed to a syntactic deficit. This possibility was

examined explicitly in Goad & White (2006). The present methodology attempts to test the

latent predictions of Goad & White’s claims across other L2 groups. If they are on the right

track, then any learner group acquiring English as an L2 whose L1 has similar phonological

constraints on complex coda formation should pattern like Mandarin natives, irrespective of

whether or not their L1 instantiates a syntactic feature for past available for transfer. The

present study investigates the interlanguage (IL) of Mandarin native speakers and compares

their performance against two other L2 English groups (L1 Japanese and L1 Spanish). As

will be seen, comparison with these two groups is crucial to support any claims that a

syntactic deficit approach over an L1 phonological transfer account for the data is privileged,

precisely because all three languages share similar phonological restrictions on complex coda

formation, despite the fact that Japanese and Spanish have the uninterpretable past [upast]

feature available for transfer. If the L1 phonology is deterministic in L2 morphological

suppliance, then the Japanese, Mandarin, and Spanish speakers, examined under the same

methodology, should evidence similar difficulties in English past tense suppliance,

specifically when the allomorph at play would be phonologically illegal in all three L1s. To

our knowledge, Spanish has never been investigated in conjunction with these two languages

to tease apart L1 phonological influence from L2 syntactic deficits.

The rest of this article is set up as follows. In section 2, we review relevant research

into generative L2 acquisition of past tense morphology and contextualize the present study

of the L2 acquisition of English past tense morphology by highly proficient adult native

5

speakers of Spanish, Mandarin, and Japanese. In section 3, for each L1 and English, we

present analyses of the syntactic representation of pastness, as well as the morphological and

prosodic representations and phonotactic constraints involved in simple past constructions.

Using these analyses as a point of departure, we detail the learning task for each L1 group,

and the predictions that fall out of RDAs versus FAAs for the different groups. In section 4,

we describe the methodology we have implemented, which include three performance tasks

that require written or oral past morphology suppliance. In sections 5 and 6, we report on our

written and oral data, respectively. As will be seen, these results cannot be easily

accommodated by RDAs, since the Mandarin and Spanish groups perform similarly across

the three tasks. These findings are unexpected given that Spanish instantiates a syntactic

feature for past, while Mandarin does not. Additionally, Japanese and Spanish speakers

perform differently, which is also unexpected given that both instantiate a syntactic feature

for past. As such, we posit that the results can be more easily accommodated by the PTH

because syntactic differences in the L2ers’ native language are not supposed to be reflected in

performance at high levels of proficiency. Section 7 addresses our findings as they relate to

the hypotheses we have tested, and considers other factors that may help account for the data.

2. Previous generative acquisition research

Two generative approaches to adult L2 acquisition of morphosyntax are tested against one

another herein: Representational Deficit Accounts (RDAs), and Full Access Approaches

(FAAs). In this section, we outline these accounts, review the most relevant studies that have

tested these approaches against one another, and contextualize the present study that follows

the same line of reasoning from a unique angle.

Representational Deficit Accounts (RDAs1) state that (some) differences at the level

of syntactic representation between the L1 and the L2 are inevitable. In its most current

1 We operationalize a group of theories under the macro-label Representational Deficient approahces

(RDA) that claim syntactic features not instantiated in the L1 grammar and thus not available for transfer are un-acquirable by second language learners in adulthood, such as Failed Functional features (Hawkins & Chan

6

instantiation, it is maintained that adult L2ers lose the ability to acquire new L2 syntactic

(uninterpretable) features from the universal inventory, whereas semantically interpretable

features remain accessible (Hawkins & Hattori 2006; Tsimpli & Dimitrakopoulou 2007).

Such approaches, however, acknowledge the possibility of L2 learning which can correspond

to what would seem to be surface reflexes of new syntactic feature acquisition, e.g., rote

learning of morphological paradigms and their application via instruction, as well as

unconscious compensatory learning strategies such as extensions of their L1 to accommodate

the L2 input.

If on the right track, one consequence of an RDA predicts the impossibility of new

uninterpretable features to be acquired in adulthood at the underlying level of representation.

Yet, if one takes seriously that L2 learning in the truest sense of the word obtains as pointed

out by Hawkins and others, then RDAs do not necessarily predict wildly different L2

performance for domains where surface learning could mask distinctions in underlying

representation, for example, in the case of overt morphological production. Rather, RDAs

expect predictable L2 variability/optionality in performance because other non-syntactic

factors can help adult L2ers perform in ways that resemble the target language, such as the

effect of explicit instruction and frequency effects. This prediction is particularly true of

several irregular past tense forms, which tend to be frequent in the input and are more salient

in the input that regular forms (e.g., ‘be’ vs. ‘was’). Therefore, RDAs would not propose a

complete lack of knowledge because L2ers would have access to other non-syntactic

resources.

The second generative SLA perspective discussed here is couched within Full Access

Approaches (FAAs), which claim that adults have access to the full set of Universal

Grammar (UG) features as in L1 acquisition (e.g., Schwartz & Sprouse 1996; Schwartz 1997) and the Interpretability Hypothesis (Tsimpli & Dimitrakopoulou, 2007). Despite differences between them, they make the same predictions as it relates to our study and thus can be treated as a single group

7

1998). FAAs propose that the noted difficulty in L2 functional morphology does not result

from inevitable target-deviant syntactic representations, and that native-like syntactic

representations are indeed attainable, especially at the advanced L2 level. However, they

argue that the complexities inherent to the task of overtly expressing functional morphology

can result in variability/optionality for a number of non-mutually exclusive reasons

(Slabakova 2008; Lardiere 2009). Importantly, these approaches must attempt to account for

the same observable optionality/variability of L2 functional morphology production in ways

that offer a tenable and testable alternative to RDAs’ claims.

Knowledge and production of L2 English past tense has featured prominently in the

debate between these two positions for L2 morphosyntax. In the remainder of this section, we

review the most relevant of these studies, which will also serve to contextualize the present

study that follows the same line of reasoning from a unique angle.

Hawkins & Liszka (2003) tested advanced L1 speakers of German, Japanese, and Mandarin

for marking of L2 English thematic verbs for simple past tense.2 Statistical analysis of data

from a sentence completion task revealed no difference among the L2er groups (χ2 = 4.94, df

= 2, p<. 05) when comparing regular and irregular verbs, and thus the authors conclude that

morphological knowledge of past inflection was alike for all L2 groups. However, data from

elicited spontaneous production tasks evidenced a significant difference between groups for

both regular (χ2 = 30.49, d f= 2, p<.01) and irregular verbs (χ2 = 8.13, d f= 2, p<.05).

Arguably, this difference was due to the Mandarin informants since their mean scores were

markedly lower, which the authors attribute to the fact that Mandarin does not instantiate a

syntactic feature for past tense, unlike German and Japanese. The researchers also explored

the possibility that L1 phonotactic transfer might have been responsible for low suppliance of

past verb morphology (as argued, e.g., in Lardiere 1998a, 1998b, 2000). They compared the

2 The details of the uninterpretable feature [upast] are provided in section 3.1.

8

performance of Mandarin and Japanese subjects on consonant clusters since both languages

impose restrictions on them in ways relevant to past tense production of certain English past

allomorphs. However, Mandarin speakers showed a markedly higher retention of consonant

clusters on monomorphemes (82%) than simple past forms (63%), which suggests that the

presence of a consonant cluster is not the determining factor in lower suppliance of past tense

morphology. However, these data also suggest that consonant clusters can be a challenge by

themselves, judging by the performance on monomorphemes. At the same time, the fact that

the Japanese speakers show a high retention rate of consonant clusters in monomorphemic

words (96%) seems to indicate that although both Mandarin and Japanese disallow consonant

clusters, somehow Japanese speakers are better able to retain them, which might be

responsible for their higher performance on simple past forms (92%).

Lardiere (1997, 1998a, 1998b, 2000, 2007) has analyzed longitudinal data from a

native speaker of Chinese (Hokkien and Mandarin), known as Patty, who is reported to

supply past tense morphology in obligatory contexts in a fairly stable manner (34.78%,

34.85%, and 33.82% at three different times over a period of nine years). The author has

argued that Patty’s command of verb morphology underrepresents her syntactic knowledge,

and that syntactic knowledge and its morphophonological reflexes need to be understood

separately (Lardiere 2000:120). It is posited that L1 phonological transfer may account for

this discrepancy, since consonant clusters are disallowed in Mandarin and Hokkien. Indeed,

Patty’s written data show at least twice as many instances of suppliance of past tense marking

in obligatory contexts than in her oral spontaneous production (Lardiere 2007), again

suggesting that non-syntactic factors can be influential in morphology suppliance. A similar

point is also made in Hawkins & Liszka (2003), although they warn that command of verb

morphology may overrepresent L2ers’ actual knowledge. We take this to mean that studies

that attempt to gather knowledge of the morphosyntactic feature for past need to consider

9

non-syntactic factors as well. Goad et al. (2003) tested 12 L1 Mandarin L2ers of English who

had resided in Canada between six months and five years. They propose that interlanguage

performance is constrained by phonological transfer effects, which can result in either across-

the-board deletion of morphophonological material or variable suppliance. In their analysis,

which we adopt in the current study, functional morphology is prosodified differently in both

Mandarin and English, which is in part responsible for variable suppliance of the past –ed

morpheme for Mandarin natives. While in English the past morpheme is adjoined to the verb

stem, in Mandarin, functional morphemes (such as the aspectual perfective marker -lɘ), are in

a position internal to the PWd, as shown in (1) (examples from Duanmu 2000:81, as cited in

Goad & White 2006:248, 251). Therefore, the learning task for Mandarin speakers consists of

not only retrieving the syntactic [upast] feature from the universal inventory, but also

prosodifying functional morphology in a new way.

(1) a. English simple past allomorph b. Mandarin aspect allomorph

PWd

PWd

Foot

σ σ

rᴂp t

PWd

Mai3 lɘ5 buy PERF

 

Goad et al. (2003) suggest that adjunction of the –ed past marker in English is

possible due to the violation of two of four constraints on prosodic domination proposed by

Selkirk (1997). The first is EXHAUSTIVITY (EXHAUST), which militates against a

constituent immediately dominating another constituent that is more than one level below it

in the prosodic hierarchy. In this case, the English prosodic structure violates

EXHAUST(PWd/ σ), which militates against a PWd linking directly to a σ. The second

violated constraint is NONRECURSIVITY (NONREC), which militates against a constituent

10

dominating another constituent of the same prosodic level (in this case, a PWd dominating a

PWd). As seen in Figure 1a, EXHAUST and NONREC are simultaneously violated in

English forms that are inflected for simple past tense, since an external PWd directly

dominates the internal PWd, while also linking directly to the external σ.3

In their study, Goad et al. (2003) propose the Prosodic Transfer Hypothesis (PTH),

which predicts that a L1 prosodic structure that is not part of the L2 grammar cannot be

acquired via access to universals. While their data support the hypothesis that transfer of L1

structures is the only possibility for suppliance, Goad & White (2004) determined that such a

position was too strong. According to the instantiation of the PTH as detailed in Goad &

White (2006), adult L2 learners have partial access to Universal Grammar for prosodic

phonology in that there is full transfer of the L1 grammar to the L2, but not full access to

universals not instantiated in the L1. Learners must therefore rely on L1 structures transferred

to the L2 to construct English past tense prosodic structure; construction of L2 structures that

cannot be built from existing L1 structures is predicted to be impossible. English learners are

predicted to prosodify inflection adjoined to the PWd in the L2 if their L1 grammar allows

the structures necessary for adjunction (recall that these include a PWd dominating a PWd

and a PWd directly dominating a σ). L2 prosodic structure construction is thought to be

possible if a) it can be built through combining L1 licensing relations or b) if it involves L1

structures being licensed in new positions (p. 247). As Goad & White (2006) state, most

languages evidence compounding and permit direct domination of a syllable by a PWd, and

therefore “learners from many L1s lacking adjunction should be in a position to build the

structure required for English-type inflection” (264).

3 While Goad and colleagues discuss acquisition of L2 prosodic structure within an Optimality

Theoretic framework consisting of constraint rankings and violations, for the purpose of this paper, we limit our discussion of the requirements for construction of new prosodic structure to the two pieces of structure needed for adjunction: a PWd linking directly to a dominating a PWd and σ.

11

Importantly, both of the structures necessary for PWd adjunction occur in Mandarin.

First, a PWd can link directly to a σ in a three-syllable construction in which the right-most

syllable is undominated by a foot in order to comply with foot binarity. Second, a PWd can

dominate a PWd in the case of compounding.4 However, these licensing relations do not

occur in a single structure, which is a requirement of adjunction. As such, Mandarin

inflectional morphology is thought to be part of the host’s PWd (what Selkirk calls an

‘internal clitic’). While adjunction is therefore not possible in Mandarin and thus condition b

of Minimal Adaptation is not met, L1 Mandarin speakers are in a position to build English

prosodic structure (although not inevitably so) because their L1 grammar includes the two

structures necessary for adjunction (Goad & White 2006). That is, condition a of Minimal

Adaptation is met.

Goad et al.’s (2003) results showed a lower rate of suppliance for regular verbs (57%)

than for irregular verbs (78%), a finding consistent with data from Hawkins & Liszka (2003)

as well as Lardiere (1998a, 1998b, 2000, 2007). These findings are in line with the

predictions of the PTH, given that pseudo-inflected forms involve organization of the past

tense morpheme inside the PWd of its host. That is, the learners are better at producing a

form whose prosodic structure mirrors their L1 structure than they are at producing the

adjoined form that is not part of their L1.

In a later study (Goad & White 2006), the authors analyzed data from 10 L1 Mandarin

speakers and nine native controls.5 Participants were required to choose (in 12 seconds) one

of two written sentences as a possible continuation to a prompt, as in the following example

(from Goad & White 2006:252):

(2) Last night after dinner…

4 See Goad et al. (2003) for details of these violations. 5 Goad and White (2006) also investigated perfective morphology in this study, which is not discussed

here.

12

-you show me photos of your daughter

-you showed me photos of your daughter

The L1 Mandarin speakers selected the correct tense 83% of the time, in comparison with the

English natives’ 98% accuracy. A two-way repeated measures ANOVA revealed a significant

main effect for group (f(1, 17) = 17.134, p< .001). However, in spite of the differences

between the two groups, the authors argue that Mandarin speakers do indeed represent [upast]

in their IL since they correctly chose past tense 83% of the time, well above chance.

Further analysis compared the “clicked” correct responses, thought to represent

comprehension, and the oral responses to further examine how oral production of a target

could be affected by phonological factors. The rate of selection of past forms in the first part

of the experiment was compared with the rate of past tense suppliance in the recorded oral

responses. In addition to examining the production of regular versus irregular forms,

production of long-stemmed (VXC-final) versus short-stemmed (VX-final) regular forms was

compared. Stem length is of interest when investigating prosodic transfer for two reasons.

The first is because learners could treat short- and long-stemmed regulars differently,

evidencing higher rates of suppliance for short-stemmed forms. This could be a possibility if

learners were to treat short-stemmed forms like pseudo-inflected forms, essentially working

around adjunction with the short-stemmed forms by constructing a PWd-internal

representation of inflection (as in Mandarin). While an English rhyme can contain a

maximum of two positions PWd-internally, a third position can occur PWd-finally. Pseudo-

inflected forms such as ‘kept’ [kεp-t]PWd are an example of this exception and are prosodified

without adjunction. Similarly, in a short-stemmed form such as [pIk-t]PWd, in comparison

with a long-stemmed VXC-final form such as [[hεlp] PWd t] PWd, the addition of the past-tense

allomorph does not exceed the maximum three-position rhyme and inflection could occur

within the PWd. The second motive for investigating stem length is the articulatory difficulty

13

that comes with a three-consonant coda cluster and implications for L1 and L2 phonotactic

constraints.

There were no statistically significant differences found between regular and irregular

forms, or between short- and long-stemmed forms, with rates of suppliance ranging from

87% (short-stemmed regulars) to 94% (ablaut). The authors present evidence of stem

reduction in the long-stemmed forms, which could suggest prosodification within a single

PWd across the board. However, a phonetic analysis of learner productions indicated

otherwise; stem-final consonants in the regular and pseudo-inflected forms were treated

differently. Thus, similarly to Goad et al. (2003), the authors argue against RDHs. However,

while Goad et al. (2003) concluded that learners would have only their L1 prosodic structure

as recourse in L2 acquisition, Goad & White indicate that learners will be able to build

adjunction structure not available in the L1 when a minimal adaptation condition is met.

The studies mentioned here differ not only in the type of data and the number of

participants (see Table 1), but also in the results obtained. This is especially true for regular

verbs, where suppliance rate goes from very high (87% in Goad et al. 2003) to very low

(45.7% in Lardiere 2007). Therefore, further examination of group performance is warranted,

especially for regular verbs, since they show greater variation even at advanced stages of

acquisition.

Table 1. Comparison of L1 Mandarin performance across four studies

Regular verbs

Irregular verbs

Number of

subjects

Type of data

Hawkins & Liszka (2003)

62.5% 84.2% 2 Oral (spontaneous)

Goad, White & Steele (2003)

57% 78% 12 Oral (picture description)

Lardiere (2007) 5.9% 45.7% 1 Oral (spontaneous) Longitudinal (9 years)

Goad & White (2006) 87% 94% 10 Oral sentence recitation (completion)

14

3. Past tense in English, Mandarin, Spanish, and Japanese

3.1. English

Our focus here is on the mental representation of the past tense in English. More specifically,

given our empirical populations, we are interested in what the syntactic featural

representation of pastness is in the IL of English as a second language by various groups of

L2ers and the extent to which this IL representation is delimited by the uninterpretable

features available from the L1 in this domain. The IL morphophonological surface forms as

well as the mapping mechanism that connect the underlying representation to the

morphological spell out are also of concern for us. Of course, pastness can be expressed in

English through adverbials or the historical present, but for ease of exposition we focus here

on the functional morpheme represented henceforth as –ed, which also has an abstract

phonological representation. As is well known, -ed affixation can manifest as non-syllabic [t]

or [d], or syllabic [ɪd], depending on the last segment of the verbal stem. In order to convey

the idea that an event or state has taken place prior to the moment of speaking, English

encodes this in the realization of the morpheme –ed adjoined to a lexical verb (e.g., walk-

walked), although some verbs will undergo suppletion (e.g., go-went), vowel change (e.g.,

run-ran) or both vowel change and affixation (e.g., buy-bought) (Lobeck 2000).

We assume that simple past morphology marking results from a series of procedures

that go from a syntactic representation to a phonological manifestation. Specifically, we

follow Adger’s (2003:166-171) analysis. The essence of his proposal is that v enters the

syntactic derivation with an unspecified uninterpretable tense feature, [uInfl : ]. T hosts an

interpretable [past] feature, which values the uninterpretable unvalued feature on v, v[uInfl : ]

à v[uInfl : past]. The tense features on T and v match, and the uninterpretable feature on v is

checked through c-command and is, therefore, deleted: v[uInfl: past].

15

Up to this point, all operations are strictly syntactic. The checked v[uInfl: past] feature

is spelled out as it interfaces with morphology, rendering it subject to the pronunciation rule

that specifies that the affix –ed (for regular verbs) be added to the stem, resulting in the

morphological representation: stem+ed. Adger describes this interface rule as in (3):6

(3) Pronounce v[uInfl : past] as ed. (Adger 2003:170)

As is easily observed, regular -ed affixation often times produces two-consonant codas, e.g.,

‘sinned’ [sɪnd], ‘fooled’ [fu:ld], ‘missed’ [mɪst], and even three-consonant codas, e.g.,

‘worked’ [wɜːrkt]. Therefore, oral production of regular past tense English forms minimally

involves instantiation of the [upast] feature and, following Goad et al.’s (2003) analysis,

attachment of the –ed morpheme to the verb stem, and in the case of pseudo inflected forms,

the past morpheme must be internal to the PWd of its host. As will be discussed in more

detail in sections 3.2, 3.3 and 3.4, Mandarin natives need to retrieve the [upast] feature from

the universal inventory, while native speakers of Spanish and Japanese can transfer it directly

from their L1. Regarding prosodification of the past morpheme, both Japanese and Spanish

natives will need to organize it in a position outside of the PWd.

3.2. Mandarin

Mandarin does not have overt morphology for expressing simple past tense (Li 1990), and

pastness is mainly indicated through the use of temporal adverbs, and/or is calculated by the

context. Therefore, the interpretation of a sentence without sufficient context or a time

adverbial is ambiguous (example from Hawkins & Liszka 2003:26):

(4) Zhangsan kan dianying

Zhangsan see-INF movie

6 For irregular cases such as ‘eat’-‘ate’, Adger proposes a special rule, expressed as: Pronounce ‘eat’ as ‘ate’ when it is adjacent to v[uInfl : past], and in this case, do not pronounce v[uInfl : past] (Adger 2003:171).

16

‘Zhangsan is seeing/saw a movie’

Since Mandarin does not select the parameterized feature [upast] from the universal inventory

(see Goad et al. 2003; Hawkins & Liszka 2003), Mandarin natives’ acquisition task involves

retrieving such feature as well as the prosodification mechanism to adjoin the –ed morpheme

to the verb host. In the case of pseudoinflected verbs, we find that, similar to Spanish and

Japanese, Mandarin disallows consonant clusters in the rhyme and even single consonants in

the rhyme are very limited. Indeed, as described in Duanmu (2007), a syllabic description of

Mandarin lists only two possible consonants in the rhyme, /ŋ/ (as in /tuŋ/ ‘winter’), and /n/ (as

in /tan/ ‘egg’). Therefore, L1 syllable transfer is of little help for native speakers of Mandarin

because the alternative of treating regular past forms as pseudo-inflected forms is contingent

on overcoming another set of L1 phonotactic constraints.

3.3. Spanish

Spanish expresses past tense through obligatory functional morphology on the verb and

therefore is assumed to instantiate the uninterpretable [upast] feature like English. However,

the Spanish affixation paradigm is much richer than in English, especially since grammatical

aspect is differentiated through separate morphemes for preterit and imperfect. Another

relevant point of comparison with English is the position that the past morpheme occupies

with respect to the PWd (the lexical verb). In order to propose a position for the past

morpheme in Spanish either inside or outside of the PWd, we examine its relationship with

stress assignment. The logic is that if the morpheme is visible to stress assignment, we can

take this to mean that it is inside the (lower) PWd. If, on the contrary, the morpheme’s

presence does not affect stress assignment, we take this to mean that it is outside of the lower

PWd. Such an approach is in line with Goad, White & Bruhn de Garavito (2011), who argue

that the Spanish plural morpheme –s is PWd-adjoined because it is invisible to stress

assignment. Here, we start with the common assumption that Spanish verb morphological

17

structure includes the following morphemes: prefix (if any), stem, theme vowel (TV),

Tense/Aspect/Mood (TAM), and Person/Number (PN) (Harris 1987; Núñez Cedeño &

Morales Front 1999; Alers-Valentín 2000, among others). Since the tense morpheme is

conflated with aspect and mood into a single complex morphological unit (TAM), we explore

the position that TAM occupies in the prosodic structure of the verb. For example, in the first

person plural past form of manejar ‘drive’, stress falls on the syllable já in the sequence

manejábamos ‘we drove’ (see (5)). This is the case for other regular verbs, in which stress is

invariably assigned before the TAM morpheme. Following Harris (1987), we assume that the

syllable following the TAM morpheme (which hosts PN) is extrametrical and is thus organized

into a higher PWd. Stress is then assigned to the word-rightmost left-dominant binary foot (a

trochaic foot), which suggests that the TAM morpheme is PWd-internal in order to ensure

that the foot is binary.

(5) PWd

PWd

Foot Foot

σ σ σ σ σ

ma ne ja ba mos

TAM PN

 

As established, knowledge of English past involves instantiation of the syntactic feature

[upast], and a morphological structure that renders stem+ed. Up to this point, the acquisition

task for a Spanish speaker is straightforward, as the difference between English and Spanish

relies only on the specified past allomorph. However, the acquisition task is challenging

18

when one considers the interface of the morphological representation with phonology. As

previously stated, English simple past forms involve PWd adjunction of the -ed morpheme,

and since the tense marker in Spanish is inside the PWd, native speakers of Spanish will

need to rearrange the L1-transferred prosodic structure to match that of the English simple

past forms.

Recall that learners are predicted to prosodify inflection adjoined to the PWd in the

L2 if their L1 a) can be built through combining L1 licensing relations or b) it involves L1

structures being licensed in new positions (Goad & White 2006:247). Spanish, Mandarin, and

Japanese are in fact all in such a position. However, Spanish speakers will have an arguably

easier task at hand, because PWd adjunction is an available prosodic structure in the Spanish

verb system that can be adapted for use in the English IL; the conflated PN morpheme is

adjoined to the PWd.7 That is, Spanish meets Goad and White’s Minimal Adaptation

condition of an L1 structure being licensed in a new position. However, just because the

structure is available does not guarantee it will be used. This study also explores the

possibility that L2ers circumvent PWd-adjunction by treating regular simple past forms as if

they were pseudo-inflected forms or monomorphemes. Therefore, we examine the restrictions

that Spanish imposes on consonants in the rhyme in word-final position, which are limited to

the following coronal consonants:

(6) a. liquids [l] [ɾ]: e.g. sal [sal] ‘salt’, mar [maɾ] ‘sea’

b. obstruents [ð̪] [s] [θ]8 e.g. red [reð̪] ‘net’, más [mas] ‘more’

c. nasal [n]: e.g. pan [pan] ‘bread’

Coda consonant clusters are mostly produced in careful speech and borrowings (see

Colina 2006:191; Piñeros 2008:109), and only the consonant /s/ can be the second element in

7 PWd-adjunction is also found in nonverbal constituents, e.g., the prefix in- is directly linked to the PWd, e.g., in-estabilidad ‘instability’ (see Peperkamp 1997:91). Likewise, PWd-σ linking also occurs with three-syllable words in Spanish, which are numerous.

8 The voiceless interdental fricative is specific to Peninsular Spanish.

19

a consonant cluster. Furthermore, in casual speech, such clusters are usually reduced by

deleting the second consonant, e.g., biceps as ['bi.sep], all of which suggests that complex

codas (such as the ones found in several simple past forms, e.g. ‘packed’) will be reduced at

least in the initial stages of English L2 acquisition.

Given the syntactic and phonological configuration of Spanish, the following

predictions can be made thus far. RDAs predict target-like syntactic representation and use

because Spanish natives are initially and subsequently aided by L1 transfer. The PTH also

predicts target-like syntactic representation via transfer, but problems in performance might

arise in oral production throughout development because past morphology in Spanish is

prosodified within the PWd. However, the PTH does not rule out the possibility that some (if

not all) advanced speakers might perform in a native-like manner with past morphology if

they are able to accommodate existing prosodic structures (i.e., PWd-PWd and PWd-σ

domination from their L1 into their L2 IL. Indeed, one limitation of the PTH is that in studies

such as this, predictions cannot be easily falsified, given that Spanish natives are in a position

to attain target-like production but are not inevitably so. On the other hand, with the data set

in this study, it is much easier to falsify hypotheses associated with RDAs: L2 speakers either

have the uninterpretable feature or not.

3.4. Japanese

Japanese expresses past tense through the allomorph –ta or –da added to the verb, as

described in Kuroda (1979), Vance (1987), and Tsujimura (2007), e.g., /kaku/ ‘write’ and

/kagu/ ‘sniff’ are the nonpast forms and their past forms are /kaita/ ‘wrote’ and /kaida/

‘sniffed’, respectively. Marking the verb with the simple past allomorph is obligatory in past

contexts, and so we follow Hawkins & Liszka’s (2003) assumption that Japanese has

instantiated the syntactic feature [upast]. As in English, the instantiation of the [upast] feature

results in the formation of a morphological rule that creates stem+ta. In terms of the prosodic

20

structure of the –ta/-da allomorph with respect to the verb, we propose that it is located

internally to the PWd, as explained below.

3.4.1. Position of the –ta/-da allomorph in the Japanese verb

In order to examine the position of the past morpheme, it is first necessary to briefly mention

the mora and pitch accent in Japanese. The mora is considered to be the appropriate level of

prosodic analysis in Japanese since it constitutes a recognizable and intuitive level of

prosodic organization for native speakers. Moras are also thought to regulate speech rhythm

and have structural implications for the PWd.9. A commonly assumed typology of language

rhythmic types classifies Japanese as a mora-timed rhythm (see Vance 2008:122). Accent in

Japanese involves pitch and verbs are said to be either accented or unaccented. Within

accented verbs, we attempt to establish the position of the –ta/-da morpheme with respect to

the verb stem by examining whether its presence is visible to pitch accent assignment

(similarly to what we have explored for Spanish in the previous section). Thus, if the past

morpheme -ta/da is visible to pitch accent assignment, this suggests that it is inside of the

PWd. If the morpheme is not visible to pitch assignment, we assume that it is outside

(adjoined to) the PWd. An examination of past forms should therefore reveal whether the

vowel in -ta/da is relevant for accent assignment. This seems to be the case since accent is

invariably assigned to the third moraic element from right to left (pitch accent is placed

before the downward arrow) in the following past forms: /maze↓ru/ ‘mix’, /ma↓zeta/ ‘mixed’,

and /aru↓ku/ ‘walk’, /aru↓ita/ ‘walked’ (examples from Vance 2008:168). This also holds in

passive and causative past forms, as can be seen in (7) (adapted from Vance 2008:169):

(7) Dictionary form /tabe↓ru/ ‘eat’ Past /ta↓beta/ Passive nonpast /taberare↓ru/

9 Moras are thought to be equal in duration. However, as pointed out in Vance (2008:117-122), experimental efforts to demonstrate this have not been successful, and results from different studies are contradictory.

21

Passive past /tabera↓reta/ Causative nonpast /tabesase↓ru/ Causative past /tabesa↓seta/ Passive causative nonpast /tabesaserare↓ru/ Passsive causative nonpast /tabesasera↓reta/

As can be seen, in all past forms pitch accent is invariably placed in the third mora to the left

and therefore we assume that Japanese past morphology is internal to the PWd because its

presence has implications for accent assignment.10

For the Japanese native, then, English simple past acquisition involves L1 transfer of

[upast], acquisition of the morphological rule stem+ed, and PWd adjunction of the English

regular past–ed allomorph, which the reader will recall requires allowance of PWd-σ and

PWd-PWd structures. This is indeed possible, as pointed out in Ito & Mester (2003). In

Japanese word clippings, syllables are linked directly to the PWd, skipping the foot. As seen

in (8), the clipped borrowing results in a PWd formed by a bimoraic foot followed by a light

syllable directly linked to the PWd.11

(8) PWd

Foot

σ σ

µ µ µ

a ni me (esyonN) ‘animation’

 

10 A reviewer, however, suggests that the Japanese past morpheme is adjoined to the PWD, positing

that “Compounded verbs can appear with tense in Japanese (unlike in Mandarin). If these compounds involve two PWds organized into a higher PWd ([[…]PWd […]PWd]PWd), then tense would need to be outside: [[[…]PWd […]PWd]PWd -ta]PWd. The implication for this study is that Japanese natives would be in a better position to produce past tense forms than native speakers of Spanish, which is very important for testing the predictions of the PTH, but not as important for the predictions of RDAs.

11 If the second syllable were heavy (i.e., two moras), it would form a bimoraic foot.

22

PWd-PWd domination also takes place in Japanese through certain compounds. Ito &

Mester (2007) discuss prosodic adjunction in Japanese compounds and propose a typology

that distinguishes between word compounds and phrasal compounds.12 Among Japanese

three-element compounds, they analyze two types of word compounds: left branching and

right-branching. In the structures they propose, a left- branching compound such as hoken-

Gaisha bÁnare ‘movement away from insurance companies’ is represented in (9) (example

from Ito & Mester 2007:103).

(9) PWd

PWd PWd

PWd PWd

hoken-gaisha bÁnare

 

This compound illustrates that structures of the type PWd-PWd are common in Japanese.

Likewise, a right branching compound such as genkin fÚri-komi ‘cash transfer’ also shows

this type of structure as shown in (10) (example from Ito & Mester 2007:103).

(10) PWd

PWd PWd

PWd PWd

genkin fÚri-komi

!

Therefore, Japanese speakers are in a position to build the required prosodic structure for –ed

inflection because there are two Japanese structures that can be combined to form it.

12 In Ito & Mester’s analysis, the number of moras of the second member of the compound determines whether a compound is a word compound or phrasal compound. If the second member of the compound has more than four moras, it becomes a phrasal compound. Otherwise, it is a word compound (see Ito & Mester 2007:104).

23

As mentioned, this study also explores whether L2ers circumvent PWd-adjunction by treating

regular simple past forms as if they were pseudo-inflected forms or monomorphemes. Thus,

it becomes necessary to examine the inventory of Japanese syllable types in order to

determine what constraints Japanese has against consonants in the rhyme. As described in

Ota (1999:24), Japanese has heavy restrictions on the presence of consonants in rhyme

position. Indeed, the only consonants allowed in the rhyme are nasal consonants, either

placeless, e.g., /hoN/, or homorganic in place with the onset of the following syllable, e.g.,

[kam.pa] ‘campaign’, [heŋ.ka] ‘change’, and the first part of a geminate,e.g., [kap.pa]

‘cucumber’, [kis.sa] ‘tea house’). Importantly, no complex codas are ever allowed and only a

placeless nasal is allowed in word final position.

Since Japanese disallows consonant clusters in the rhyme, this means that L1 syllable

transfer is of little help for both regular simple past forms and pseudo-inflected forms.

Indeed, as is usually observed, English borrowings in Japanese are fixed to conform to L1

phonotactic constraints. The phonological acquisition task for native speakers of Japanese

thus involves building the prosodic structure for regular past forms and/or allowing consonant

clusters for pseudo-inflected forms.

3.5. Interim summary: Predictions and hypotheses

As discussed in detail above, Mandarin, Spanish, and Japanese disallow coda clusters (in

similar ways) and have the necessary elements of prosodic structure for adjunction in their

grammars, only Mandarin lacks the uninterpretable [upast] feature, and therefore a

comparison of past suppliance among these groups should shed light on the role that lacking

[upast] alone may have on past morphology suppliance. Thus, RDAs predict that Mandarin

natives will consistently perform below native speaker level, differently (i.e., worse) than

Spanish and Japanese natives. It is also predicted that there will be no observable differences

between oral and written responses, provided they are appropriately comparable. On the other

24

hand, the PTH (and other FA approaches), predicts that such oral-written asymmetry among

these groups may simply not take place. Instead, all three groups are predicted to perform in a

native-like fashion when they produce past forms in writing. The PTH suggests that a great

deal of variation might occur when responses are orally produced, however, irrespective of

whether the L1 can provide the underlying syntactic feature relevant for the L2 category,

since prosodic structure for simple past in each language is not created via adjunction.

4. The Study

4.1. Subject population

The experiment had four groups: a native English control group (n = 15), and three groups of

L2ers whose native languages were Mandarin (n = 15), Spanish (n = 13), and Japanese (n =

11). All participants were recruited in academic settings in different areas of the United

States, and were first exposed to daily interaction with English native speakers at around or

after age 18. Exposure to naturalistic input was deemed important to establish that the

participants had had enough evidence that past tense morphology is overtly manifested in

lexical verbs. Each participant had been living in the US for a minimum of one year four

months up to 25 years, and none was proficient in a third (or subsequent) language. A similar

protocol as the one in Hawkins & Liszka (2003) was used to determine proficiency in

English. That is, subjects were selected on the combined score obtained in the multiple-

choice grammar test and the multiple-choice vocabulary test of the Michigan Placement test,

and only participants whose scores on both measures fell around or above 75% accuracy were

selected.13 Table 2 shows the L2ers’ profiles for Age of arrival (AoA), length of residence

13 Hawkins & Liszka used Nation’s (1990) vocabulary test at the 10,000 word level. However, the

Grammar and Vocabulary section of the Michigan Placement test was deemed more appropriate, since the learners had been exposed primarily to American English. The cutoff in their study was 80%, but we chose 75% was an appropriate cutoff because most scores clustered between 75% and 85%. Additionally, a closer examination of the combined proficiency scores and the performance on past tense suppliance did not suggest any correlation between higher proficiency scores and higher performance. For example, in all L2er groups,

25

(LoR), and combined proficiency score (Prof). Within each language group, each row

represents a participant. All scores have been organized according to proficiency scores, with

the highest scores at the top of the table and the lowest scores at the bottom. The Spanish

group had the highest proficiency score (M = 86.76, SD = 6.45), followed by the Japanese

group (M = 83, SD = 7.5), and the Mandarin group (M = 79.4, SD = 4.2). A univariate

ANOVA indicated a significant difference in mean proficiency score among groups (F(2,39)

= 5.193; p = .01). Bonferroni post-hoc tests revealed that the Mandarin group had

significantly lower scores than the Spanish group (p = .008), but no other group differences

were significant.

Table 2. L2ers’ length of residence and proficiency score

Japanese Spanish Mandarin

AoA LoR

Prof AoA LoR

Prof AoA LoR

Prof

26 24 98 27 5;4 98 25 1;4 87 27 13 92 24 4;9 93 22 1;6 86 28 21 90 24 17 93 24 1;4 85 27 11 85 25 8 91 22 3;4 82 19 11 84 32 10 91 17 7 81 21 13 80 26 1;6 89 26 6;6 81 28 1;4 78 28 8 88 23 1;6 79 29 5 77 32 2;4 84 22 2;6 77 19 9 77 27 14 84 20 3;4 77 19 9 76 36 4 82 29 1;7 76 27 1;4 76 42 8;4 79 23 1;4 76

30 4;6 79 25 9 75 25 6 77 23 2;6 75 26 1;4 74

The control group was composed of native speakers of American English (n = 15) studying at

different universities in the US. Their combined proficiency scores ranged between 80% and

100% (M = 95.07, SD = 5.75), and their age range was between 18 and 31 (M = 21.06, SD =

3.17). some participants with proficiency scores between 75% and 80% outperformed participants with proficiency scores of 90% or higher.

26

4.2. The tasks

Three tasks were carried out: two sentence completion tasks and a picture description task.

While the sentence completion tests provide a controlled measure of use of functional

morphology, the picture description task aimed at obtaining a more natural sample of past

tense morphology use. All oral responses were recorded using a headset connected to a

computer (using PRAAT software, Boersma & Weenink 2014, and were broadly transcribed

by two trained native speakers of English.

4.2.1. Sentence completion tasks

Each participant completed two sentence completion tasks. In the first (Task A), they read 82

short contexts (each controlled to be 20-25 syllables long) in which one word was missing,

and they were asked to write the missing word to complete the sentence.14 In the second

sentence completion task (Task B), they read aloud 82 short contexts (20-25 syllables long)

and were asked to utter the missing word out loud.15 These two sentence completion tasks

had the same design and therefore provide highly comparable data: the same targets, number

of sentences, and 16 distracters. The tasks were also completed under the same conditions:

participants were told that there was no time limit and that they could change their answers if

they wanted. The only difference in the design was the context in which the targets were

embedded in order to avoid priming effects. The contexts consisted of one or two sentences

taken from corpora so that they would resemble the kind of input that L2ers are typically

exposed to and minimally adapted to control the environment. For example, for the target

‘filled’, two contexts were generated:

14 Targets were placed at the right edge of the sentence, and were either in final position or right

before a word with a vocalic onset in order to facilitate data coding.

15 One reviewer inquired as to why only one word in each set could fit in each blank. The task was designed in this manner to ensure that participants would all choose the same answer and therefore the data would be comparable across participants and tasks.

27

(11) a. Johnny had a terrible headache, so he _______ a glass with water and took two

aspirins. [fill write type]

b. In 1950, all computers in existence were huge. Each one _______ a whole room!

[send fill smell]

To control for task effects, approximately half of the participants in each group

provided oral responses to Task A, and written responses to Task B, while the other half

provided written responses to Task A, and oral responses to Task B.

4.2.1.1. Targets: Simple past forms and monomorphemes

A total of 27 simple past forms were included in tasks A and B. Of these, 15 were regular

simple past forms with a complex coda, six with a singleton coda, and six were irregular

(pseudo-inflected) forms. Fifteen monomorphemes that were similar in phonological shape to

regular simple past forms were included as well (e.g., ‘past’, see Table 3). This was in order

to obtain an indication of whether phonological constraints affected consonant clusters across

the board in the IL. This is particularly relevant given that Mandarin, Spanish, and Japanese

disallow consonant clusters often seen in English simple past forms and lexical

monomorphemic words. Since the goal was to have comparable sets of past forms and

monomorphemes, only past forms whose consonant cluster had a similar monomorphemic

counterpart were included. This resulted in a selection of 15 past forms and 15

monomorphemes that ended in the consonant clusters [ld], [nd], and [st]. Additionally, in

order to obtain a fairly uniform selection of targets, all past tense forms and monomorphemes

were monosyllabic. To control for frequency, all targets were selected from the 5,000 most

common words in the Corpus of Contemporary American English (COCA) and we followed

a protocol similar to that in Marshall & van der Lely (2006). That is, for every token, we

obtained the total number of hits from the COCA, and then calculated the natural logarithm

28

for each value.16 Taking into consideration the natural log values, we ran a t-test between the

past forms and the monomorphemes (α = .05) to verify that there were no statistically

significant differences between each set. The resulting list was: ‘build’, ‘field’, ‘wild’, ‘mild’,

‘bald’, ‘send’, ‘blind’, ‘pound’, ‘blend’, ‘pond’, ‘test’, ‘list’, ‘dust’, priest’, and ‘nest’ (see

Appendix 3 for complete information).

Simple past forms with a single consonant in the rhyme were originally included (n =

6) as well as pseudo inflected forms (n = 6). This helped gather data that were more inclusive

of simple past forms in general, providing a more global picture of overall performance on

simple past use.17 All verbs included in this condition were monosyllables that resulted in a

three-element rhyme with the single consonant [d], resulting in the tokens ‘played’, ‘fried’,

‘tried’, ‘paid’, ‘cried’, and ‘stayed’. Previous research on L2 English past tense has shown

that irregulars are supplied differently than regular verbs by L2ers. This asymmetry has also

been a matter of discussion in terms of whether regular and irregular verbs are represented in

the mind in the same way (see Pinker 1991; Clahsen & Rothweiler 1992; Prasada & Pinker

1993; Bybee 1995; Clahsen 1999; McClelland & Patterson 2002; Pinker & Ullman 2002).

For example, Lardiere (2007) noticed that in Patty’s data, irregulars were more often marked

with past morphology than regular verbs, but that even though the most frequent verb used

was have, it was never marked for past. Additionally, Patty’s written data show almost no

difference in suppliance of past verb forms between regular and irregular verbs. Hawkins &

16 Calculation of the natural logarithm of the total number of hits makes comparisons easier, given that

the values range greatly, e.g., ‘killed’ had 46,714 hits and ‘field’ had 72,594 hits. Their natural logarithms, however, are 10.75 and 11.2, respectively.

17 We do not report results about these two types of verb for a number of reasons. The number of regular simple past forms was much higher than either pseudo inflected (n = 6) or singleton coda past forms (n = 6), additionally, singleton coda past forms and pseudo inflected forms were much more frequent in the corpus than regular simple past forms, making comparisons between these conditions was less than ideal. The reason why fewer tokens in these two conditions were included was that a larger pool would have meant an even greater difference in terms of frequency. It was also found in the results that some targets were frequently omitted by participants (specifically, ‘kept’ and ‘went’), making the comparison even weaker.

29

Liszka (2003) found higher accuracy on irregulars, while Goad & White (2006) found no

significant difference. The past irregular forms selected for inclusion had to end in a

consonant cluster and be distinguishable from their infinitival forms phonologically and

orthographically. As such, pairs such as ‘cut’ – ‘cut’ were not considered. The complete list

of past forms used in the tasks is outlined in Table 4.

Table 3. Past tense tokens

sealed, filled, killed, yelled, rolled Regular simple past forms with consonant clusters joined, cleaned, trained, ruined, gained

passed, pressed, kissed, blessed, missed Regular simple past forms with singleton coda consonant played, fried, tried, paid, cried, stayed

Pseudo inflected forms found, told, kept, left, went, lost

4.2.2. Picture description task

The oral picture description task was included to examine whether the patterns observed in

the sentence completion tasks would hold in a task in which participants have more control

over the context, the vocabulary, and the structures used. It was also the expectation that

subjects’ attentional resources would not be as focused on the targets, thus reducing the

possibility of subjects using or relying on metalinguistic knowledge.18 In order to elicit the

expected simple past targets, each picture contained printed words or short phrases that called

for the intended targets. Participants were first asked to describe two pictures as practice.

Afterwards, they were introduced to a fictional character, Jane, and were asked to describe a

typical day of work for Jane by looking at a series of pictures of a woman performing

different activities. Then, they were given a series of images that showed Jane’s day off,

which was a set of pictures that elicited simple past tense (see Appendix 1). In order to set a

context of pastness, participants were told the following:

18 See Bonner (2013) for details on effects of performance factors on oral suppliance.

30

OK, we know Jane, and we know she is a secretary and that she works a lot. In fact,

she works too much, but she finally took a day off, and so we are interested in talking

about that. Here are all the things that she did yesterday. So, just like before, describe

what she did and follow the sequence of the pictures. Make sure you use the words

below each picture and mention the times.

In order to ensure that participants would interpret all the events as taking place in the

past, rather than as habitual actions or ongoing events, they were asked to start the description

with the prompt ‘yesterday, it was Jane’s day off...’. The targeted past forms were: ‘slept’,

‘worked’, ‘bought’, ‘washed’, ‘cleaned’, ‘baked’, ‘walked’, ‘drank’, and ‘went’.

4.2.3. Data coding

For all tests, past morphology suppliance was coded in the following way. In a target such as

‘called’, if the second consonant of the consonant cluster [ld] was not pronounced, this was

considered not to be supplied. Conversely, if the consonant that carries past tense

morphology (here, [d]) was pronounced, it was considered as supplied. Past tense

morphology was also considered supplied in the case that the stem-final consonant was

deleted but the tense morpheme was supplied (e.g., [kli:d] for ‘cleaned’). Likewise, in

comparing simple past forms against monomorphemes, when the second consonant of the

cluster in the monomorphemes was not pronounced, it was coded as not supplied, e.g., ‘cold’

pronounced as [koul]. The idea was that, had it been a simple past form, it would have been

interpreted by a potential listener as an infinitival form. For each participant in each group, a

measure was obtained for his/her overall past suppliance.19

19 In the picture description task, a total of three native, two L1 Spanish, three L1 Mandarin, and four

L1 Japanese productions were eliminated from the count due to potential coarticulation effects, e.g., if a participant said ‘washed the car’, it was difficult to determine whether the past tense morpheme was actually being produced since the onset of the following word was /ð/.

31

5. Results: Sentence completion tests

5.1. Between-group comparisons: Written responses

Figure 1 shows suppliance of simple past morphology in all written data. An ANOVA was

conducted, and since the assumption of homogeneity of variance was violated; the Welch F-

ratio is reported, revealing a non-significant main effect F(3, 23.915) = 1.539, p >.05.

Therefore, differences among groups are not statistically significant, and the groups evidence

high levels of suppliance. This finding suggests that when morphology suppliance is not

dependent on phonological production, all groups perform in a native-like fashion regardless

of the L1 feature configuration. We take this evidence to not support the predictions of

RDAs, although we acknowledge that these approaches do not preclude metalinguistic

knowledge and/or other compensatory strategies from explaining this better performance in a

written modality.

Figure 1. Simple past suppliance in written responses

We also examined L2er individual performance in order to quantify how many of the

L2ers truly performed in a native-like fashion. A look at the control group data shows that

controls performed within a range from 92% to 100%, which therefore constitutes the native

98.98 97.95 97.5 94.08

0

20

40

60

80

100

Native controls (n=15)

L1 Spanish (n=13)

L1 Mandarin (n=15)

L1 Japanese (n=11)

32

speaker range. In both the Spanish (12 of 13) and the Mandarin (14 of 15) group, all but one

subject in each group performed within the native range, and in the Japanese group, eight out

of 11 did. These data suggest that when morphology is supplied in written modalities, most

advanced speakers perform in a native-like way regardless of the presence of [upast] feature

in their L1.

5.2. Inter-group comparisons: Oral responses

This data set contains all (oral) simple past forms in the study: regulars with a consonant

cluster coda, regulars with a singleton coda, and irregulars with consonant clusters. As shown

in figure (2), native controls supplied past morphology close to ceiling (M = 96.7%, SD =

3.39), while the Spanish group provided past morphology at a lower rate (M = 83.24, SD =

15.26), as well as the Mandarin group (M = 79.49, SD = 12.12) and the Japanese group (M =

88.41, SD = 17.37).

Figure 2. Simple past suppliance in oral responses

A one-way ANOVA was conducted and since the assumption of homogeneity of

variance was violated, the Welch F-ratio is reported, revealing a significant main effect for

Group (F(3, 21.217) = 11.9, p< .001). The Games-Howell post hoc test reveals that the

96.78

83.24 79.49 88.41

0

20

40

60

80

100

Native controls (n=15)

L1 Spanish (n=13)

L1 Mandarin (n=15)

L1 Japanese (n=11)

33

Native control group’s rate of suppliance is significantly higher than both the Spanish group

(p = .049) and the Mandarin group (p = .000), but no significantly different from the Japanese

group (p = .315). Among the L2er groups, no significant differences were found (p > .05).

The data here support some of the predictions: the Mandarin group performs differently than

the control group, and the Japanese group performs in a native-like fashion. However, the

Spanish group data cast some doubts since RDAs predict that performance should be target-

like for this group, as well, and certainly that this group should fare no worse than the

Japanese group. Recall that we included the Spanish group precisely to have a novel group in

the literature that unambiguously has the relevant syntactic features in the L1 but similar L1

prosodic constraints to Mandarin and Japanese, which are comparably much more widely

studied. For the PTH, the predictions leave more room for variation since it stipulates that

target-like performance in oral data is possible if the target prosodic structure can be built by

combining existing structures (which is the case in all of the L1s at play).

Since the PTH claims that some L2ers may attain a target-like prosodic

representation, it is worth examining individual data. The control group performed within a

range of 88% to 100% suppliance, which constitutes the native speaker range. In the Spanish

group, six out of 13 performed within the native range; in the Mandarin group, six out of 15

were within the native range; in the Japanese group, nine out of 11 were within the native

range. As can be seen, the data (individual and aggregate) pose a challenge for RDAs, given

the difference between the Spanish and the Japanese group, on the one hand, and the fact that

a third of the native speakers of Mandarin perform within the native range. On the other hand,

this data set does not pose any serious challenge for the PTH, given that native-like

performance is a possible (although not inevitable) outcome.

34

5.3. Native control group data

5.3.1. Written and oral performance

A comparison between written and oral data for all groups shows the extent to which

phonological factors can affect morphology suppliance. As expected, past morphology

suppliance is very high in both modalities for the control group, around 97%-98%, as shown

in figure 3. These scores were submitted to a Wilcoxon signed-rank test, revealing that past

suppliance in written responses (M = 98.98, SD = 2.27, Mdn = 100) was not significantly

higher than past suppliance in oral responses (M = 96.78, SD = 3.39, Mdn = 96.29), T = 14.5,

p>.05 (see Figure 3).

Figure 3. Simple past suppliance (written and oral responses) in control group

5.3.2 Consonant cluster effects on simple past forms

Since it is possible that L2ers of English resort to different strategies to accommodate for

inflection, especially if their L1 disallows coda consonant clusters, consonant clusters in three

conditions were examined:

96.78 98.98

0

20

40

60

80

100

Past (oral) Past (written)

35

(i) Monomorphemes with consonant clusters [ld], [nd], and [st] (oral

responses)20

(ii) Simple past forms with consonant clusters [ld], [nd], and [st] (oral responses)

(iii) Simple past forms with consonant clusters [ld], [nd], and [st] (written

responses)

Figure 4 shows the controls’ performance in the three aforementioned conditions. A

repeated-measures ANOVA was carried out and a Mauchly’s test indicated that the

assumption of sphericity had been met (χ2(2) = 1.3, p>.05), showing that the overall rate of

suppliance was not affected by the type of answer, F(3, 50) = 2.25, p >.05. This verifies that

the type of production (consonant clusters in monomorphemes, oral past forms, and written

past forms) is not a factor in native speakers’ suppliance, and that there are no significant

differences in consonant cluster production when they involve simple past morphology in

oral production (M = 97.33, SD = 4.91), simple past morphology in written production (M =

98.66, SD = 3.73), or when they are part of a monomorpheme (M = 98.22, SD = 3.05).

20 Recall that monomorphemes were coded as if they were simple past forms, e.g., if the word nest

[ˈnest] was uttered as [ˈnes], it was coded as a ‘miss’.

98.22 97.33 98.66

0

20

40

60

80

100

Monomorphemes with CC clusters (oral)

Simple past forms with CC clusters (oral)

Simple past forms with clusters (written)

36

Figure 4. Simple past and monomorphemes in control group

5.4. Comparisons within the Spanish group

5.4.1. Written vs. oral performance

The data from the Spanish group were submitted to a Wilcoxon signed-rank test revealing

that simple past suppliance in oral responses (M = 83.24, SD = 15.26, Mdn = 81.48) was

significantly different from simple past suppliance in written responses (M = 97.95, SD =

4.32, Mdn = 100), T = 2, p <.05 which strongly suggests that performance is affected by

factors other than the L1 featural configuration, namely, phonological factors. Since RDAs

do not, in principle, predict differences between performance in written and oral responses,

this data set constitutes a challenge for it. On the other hand, the PTH predicts the possibility

of markedly better performance on written responses, and therefore it seems better equipped

to deal with such differences.

Figure 5. Past tense suppliance (oral and written responses)

83.24

97.95

0

20

40

60

80

100

Simple past (oral) Simple past (written)

37

We interpret these results as evidence that when Spanish natives need to orally produce

simple past forms, their performance is hindered by phonological factors, arguably the

articulation of complex codas inexistent in their L1. By extension, then, the same could be

true of other L2er groups whose L1 do not have complex codas.

5.4.2. Consonant cluster effect on simple past forms

The results in Figure 6 were submitted to a repeated-measures ANOVA. Mauchly’s test

indicated that the assumption of sphericity had been met (χ2(2) = 1.3, p > .05), showing that

the overall rate of suppliance was affected by the type of answer, F(2, 24) = 7.57, p < .05).

This was followed by tests of Within-Subjects Contrasts, showing no significant contrast

between monomorphemes with consonant clusters (oral) (M = 74.79, SD = 20.16), and simple

past forms with consonant clusters (M = 76.63, SD = 21.83), F(1, 12) = .06, p>.05. However,

a significant contrast was observed between simple past forms with CC clusters (oral) and

simple past forms with CC clusters (written) (M = 97.39, SD = 4.37), F(1,12) = 13.788, p <

.05.

74.79 76.63

97.39

0

20

40

60

80

100

Monomorphemes with CC clusters (oral)

Simple past with CC clusters (oral)

Simple past with CC clusters (written)

38

Figure 6. Past tense and monomorphemes with CC clusters: Spanish group

We interpret these results as supporting the claim that when L1 Spanish speakers do

not simple past morphology is not mediated via phonological production, their performance

is native-like. However, when past morphology is mediated through phonology, performance

is significantly lowered. Furthermore, since monomorphemes and simple past forms are

supplied at the same (below native level) rates, we take this as suggestive evidence that L1

Spanish speakers treat consonant clusters in simple past forms the same as consonant clusters

in monomorphemes. In other words, the –ed morpheme is prosodified internally to the PWd

and, in a number of cases, subsequently deleted from the output. The relevance of this finding

is that the presence of the uninterpretable past feature is clearly not enough for Spanish

natives to supply simple past morphology, and that consonant cluster reduction by itself

shows to be a determinant factor in production of simple past forms. Consequently,

presence/absence of the uninterpretable feature in the L1 is not a sufficient condition for

native-like performance even at advanced stages of acquisition, which implies that when

Mandarin speakers fail to produce past forms in a native-like fashion, they may well be

affected by phonological factors, just as the Spanish natives show.

To sum up, the Spanish group shows native-like use of past forms in written

responses, but target-deviant behavior in oral responses. Given that their performance is

lower in oral responses, we propose that L1 phonological transfer is affecting performance.

To test this possibility, further examination points to L1 Spanish speakers having a single

phonological structure assigned to monomorphemes with consonant clusters and simple past

forms with consonant clusters. Because the L1 Spanish speakers treat consonant clusters in

monomorphemes and simple past forms in a similar (target-deviant) manner, and show a

higher rate of suppliance in past forms with a singleton coda, we can conclude that the

morphological production differences between this group and the controls is mainly a

39

phonological one. Recall that Spanish has the relevant syntactic feature for past instantiated

and it is thus available for transfer. RDAs do not, therefore, anticipate that these learners

should have deficits in underlying representation for past in L2 English (nor does the PTH, of

course). These data alone do not present any direct challenge to RDAs, but they allow us to

isolate how phonological factors weigh on functional morphology. Additionally, this

discussion will be crucial to the development of the logic we present in section 5.6 as it

relates to the Mandarin group for which other studies have claimed their suppliance of past

tense morphology in L2 English supports representational deficit L2 theories. If we show the

same for the Mandarin group as we show here for the Spanish group, then we will by

extension be able to argue that these and previous data on past morphological suppliance

cannot be used to unambiguously support RDAs, but might in fact be more accurately

understood as showing an L1 intervention effect at the level of phonology.

5.5. Comparisons within the Japanese group

5.5.1. Written vs. oral performance

Figure 7 shows the Japanese group’s scores for these two conditions. Simple past suppliance

was high for both oral responses (M = 88.41, SD = 17.37, Mdn = 95.83) and written

responses (M = 94.08, SD = 8.48, Mdn = 100). The results from a Wilcoxon signed-rank test

revealed that suppliance in simple past forms in oral answers was not significantly higher

than for simple past forms in written answers (T = 11, p>.05). These results are in line with

the expectations of RDAs for this group in that they show that L1 Japanese speakers perform

in a native-like fashion across modalities. The results can also be viewed as consistent with

the PTH, which argues for the possibility of target-like prosodic structures being built in the

IL.

40

Figure 17. Past tense suppliance (oral and written responses): L1 Japanese group

5.5.2 Consonant cluster effects on simple past forms

The same three conditions were examined as with the control and Spanish groups.

The results showed high rates of suppliance for all conditions: monomorphemes with CC

clusters (M = 94.54, SD = 12.22); simple past forms with CC clusters (oral) (M = 83.22, SD =

21.86); simple past forms with CC clusters (written) (M = 92.72, SD = 11.72). A repeated-

measures ANOVA was carried out and Mauchly’s test indicated that the assumption of

sphericity had been met (χ2(2) = 3.55, p > .05). The results showed that the overall rate of

suppliance was not affected by the type of answer, F(2, 20 ) = 1.698, p > .05).

88.41 94.08

0

20

40

60

80

100

Simple past (oral) Simple past (written)

41

Figure 8. Past tense and monomorphemes: L1 Japanese group

Thus, these results show that L1 Japanese speakers’ morphology suppliance is not

significantly affected by phonological factors, while also showing high levels of suppliance.

This data set is consistent with the predictions of RDAs because Japanese has instantiated the

[upast] feature in L1 acquisition, as well as the PTH, in the sense that it is possible for

Japanese natives to build the target prosodic structures for inflectional morphology. It is

intriguing that Japanese speakers perform significantly better than the Spanish natives in the

oral responses, given that both languages face similar phonological challenges. A possible

explanation could be that as a group, these L1 Japanese participants had a longer length of

residence (LoR) in the US than the Spanish group. If LoR differences are indeed a factor, this

would cast doubts on our results, since simple past suppliance could be attributed to

proficiency rather than common linguistic properties between the L1 and the target

94.54

83.82

92.72

0

20

40

60

80

100

Monomorphemes with CC clusters (oral)

Simple past with CC clusters (oral)

Simple past with CC clusters (written)

42

language.21 One way to examine this possibility is to examine individual demographic

information and data from each L2er, which we take up in section 6.2.

5.6. Comparisons within the Mandarin group

5.6.1. Written vs. oral performance

As can be seen in Figure 9, there is an asymmetry between the rates of suppliance for the

written and oral conditions. A Wilcoxon signed-rank test showed that written suppliance of

past forms (M = 97.5, SD = 3.64, Mdn = 100) was significantly higher than oral suppliance of

simple past (M = 79.49, SD = 12.12, Mdn = 80.76), T = 0, p >.05.

Figure 9. L1 Mandarin oral and written past tense morphology suppliance

The results from these two conditions are not predicted by RDAs in two ways: First,

performance on the written responses is native-like, and second, the performance between

21 It should be noted that all efforts were made to pair L2ers according to their level of proficiency, age of arrival and length of residence. However, difficulty in procuring L1 Japanese participants made it a challenge to control for these variables.

79.49

97.5

0

20

40

60

80

100

Simple past (oral) Simple past (written)

43

these two conditions is markedly different. Since RDAs do not incorporate phonological

factors in morphology suppliance, this is hard to explain under an RDA, although Hawkins &

Liszka (2003) examine the effect of L1 phonotactic constraints in morphology suppliance and

find that consonant clusters did not appear to be a factor responsible for lower morphological

suppliance among the Mandarin speakers. On the other hand, for the PTH, the difference

between oral and written responses is not unexpected, since phonological factors are thought

to influence morphology suppliance.

5.6.2. Consonant cluster effects on simple past forms

As before, the following comparisons were made: Monomorphemes versus consonant

clusters, simple past forms versus consonant clusters (oral suppliance), and simple past forms

versus consonant clusters (written suppliance); results are illustrated in Figure 10. A

repeated-measures ANOVA was carried out and Mauchly’s test indicated that the assumption

of sphericity had been met (χ2(2) = 1.8, p >. 05), showing that the overall rate of suppliance

was affected by the type of answer, F(2, 28 ) = 31.55, p < .05. This was followed by tests of

Within-Subjects Contrasts, showing a significant contrast between monomorphemes with

consonant clusters and simple past forms with consonant clusters (oral), unlike the L1

Japanese and Spanish groups, F(1, 14) = 16.1, p< .05. A significant contrast was also

observed between simple past forms with consonant clusters (oral) and simple past forms

with consonant clusters (written), F(1,14) = 33.44, p< 05. In summary, all visual differences

in Figure 10 proved to be statistically significant.

44

Figure 10. L1 Mandarin inflected forms versus monomorphemes

The findings that 1) oral rate of suppliance is lower than written rate of suppliance

and 2) learners are better at producing complex codas in monomorphemes than in inflected

forms strongly suggest that when L1 Mandarin speakers do not need to deal with

phonological factors, their performance is native-like, but when past morphology is mediated

through phonology, performance is negatively affected. The higher suppliance in CC

monomorphemes compared with inflected CC forms also points to the possibility that the

inflection is prosodified in an adjoined structure. The fact that monomorphemes show a

significant (but not absolute) degree of consonant cluster reduction precisely shows that

consonant clusters are challenging by themselves, and therefore, production of simple past

forms will imply violation of L1 phonological restrictions. In order to further examine

whether or how phonological factors affect morphology suppliance (and the implications for

how these data support RDAs or the PTH), in the next section we compare past morphology

production between targets that resulted in simple and complex codas. The logic behind

doing so is that if the problem in morphology suppliance stems from a syntactic deficit only,

87.7

71.49

96.4

0

20

40

60

80

100

Monomorphemes with CC clusters (oral)

Simple past with CC clusters (oral)

Simple past with CC clusters (written)

45

then the rates of suppliance for simple coda past forms should not be significantly different

from suppliance of past forms in complex coda past forms.

5.7. Summary of results from sentence completion tasks

As seen, neither RDAs nor the PTH can fully account for the data sets presented here.

However, some generalizations can be made about the acquisition of morphosyntax and its

connection to L1 phonological influence as it relates to morphological production. First, the

preponderance of the data shows that differences between morphology supplied in written

and oral modalities can be marked, and L1 phonological factors can (negatively) influence

(functional) morphology suppliance. We take this to mean that an examination of L1

phonological factors in L2 oral performance is warranted in general. The Spanish group data

and their striking similarity to the Mandarin group data underscore that the possibility of L1

transfer of target syntactic features does not nullify L1 phonological effects. By extension, if

L1 phonological effects are likely responsible for the behavior of the Spanish group, and

given the striking conformity of the Spanish and Mandarin group performances, it seems

reasonable to suggest that these target-deviant performances stem from the same source. That

is, these performances are rooted in L1 transfer at the morphosyntax-phonology interface. At

this point, we can make the following generalizations about the data in connection to the

predictions of RDAs and the PTH:

(i) Phonology is an undeniable factor in L2 morphology suppliance, especially as

evidenced by the data discussed for the L1 Spanish and the L1 Mandarin

groups, given that performance in written responses is higher than in oral

responses. This asymmetry suggests that phonological factors strongly

influence morphology suppliance and that they may be much more complex

than they appear.

46

(ii) In written performance, all L2er groups perform in a native-like fashion, and

the great majority of individual speakers perform within the native range.

(iii) Only the L1 Japanese group performs in ways similar to the control group in

all conditions. RDAs are seemingly supported by the results from the Japanese

group, the target-deviant oral performance of the Mandarin speakers, and the

target-like written performance of the Spanish speakers. However, RDAs

cannot easily account for the target-like written performance of the Mandarin

group and their high performance on past forms with a single consonant. The

target-deviant oral performance of the Spanish group also presents real

challenges for RDAs, at least in isolation of considering phonological factors.

The PTH is seemingly supported by the difference in performance observed in

both the Spanish and Mandarin group between the written and the oral data. Additional

(although indirect) evidence for a phonological account comes from the trend observed in

the L2er groups of reducing consonant clusters in monomorphemes and arguably in past

forms. Unlike the Mandarin and Spanish speakers, however, recall that that the Japanese

learners as a group are much more successful in acquiring the L2 phonological structure

than the other two groups. 22

6. Picture description task results

As can be recalled, this study includes a picture description task in order to elicit oral

production of past morphology. In what follows, we report the results and compare them with

data from the sentence completion tasks.23

22 As one reviewer points out, if the tense morpheme is represented similarly in English and

Japanese, then these results are not surprising. This is because the learning task for the Japanese speakers is confined to syllable structure complexity and is therefore simpler.

23 Auxiliary forms and the copula be were removed from the count.

47

6.1. Group results

The first comparison across groups provides the rates of past morphology suppliance (Figure

11). All controls performed at ceiling (M = 100, SD = 0), the L1 Spanish group performed

highly (M = 86.9, SD = 17), similarly to the L1 Mandarin group (M = 87.6, SD = 16) and the

L1 Japanese group (M = 90.9, SD = 15.9). While a Kruskal-Wallis test revealed a significant

effect of group on past morphology suppliance (χ2(3) = 9.226, p < .05), such an effect is most

likely due to the control group, given that its mean and standard deviation are markedly

different from those of any of the L2er groups.24 This test was followed by a one-way

ANOVA run only on the learner data. Homogeneity of variance was met (χ2(2, 34) = .003, p

> .05), and the results revealed no significant differences among the groups, (F(2, 36) = .496,

p > .05). Since the L2er groups perform similarly to one another, this data set suggests that

instantiation of the [upast] in the L1 does not seem to affect suppliance of simple past

morphology.

Figure 11. Simple past production in picture description task

24 An ANOVA could not be carried out because homogeneity of variance was violated and robust tests

of equality of means could not be performed for accuracy because the control group’s variance was zero.

100

86.9 87.6 90.9

0

20

40

60

80

100

Native controls (n=15)

L1 Spanish (n=13)

L1 Mandarin (n=15)

L1 Japanese (n=11)

48

Thus, at this point we can conclude that at advanced stages of development, L2ers’

morphology suppliance does not seem to be inevitably limited by the L1 syntactic feature

configuration (in this case, the [upast] feature), since the Mandarin group performs similarly

to the Spanish and Japanese group. However, this data set also suggests that L2ers and the

native controls perform differently in less controlled types of test, and that phonological

factors are most likely at play given that performance on written responses was target-like for

all L2er groups. It becomes relevant, then, to examine individual performance in order to

examine how many participants (and in what way) perform in a native-like way, and to

explore other factors that could be at play. We examine individual variation in the next

section.

6.2. Individual performance

Given that the results of the elicited production task are different from the sentence

completion tests, it is worth exploring native-like performance in individual L2ers. Recall

that RDAs make strong as well as compelling claims for L2 acquisition: New L2 syntactic

features are inaccessible to adults as a result of some type of biological maturation, which is

an appealing idea given that generative research assumes that language is a biological

phenomenon. If this is indeed the case, then, the Japanese and Spanish group should pattern

together, while the Mandarin group should be consistently different from both of them.

Likewise, most if not all native speakers of Japanese and Spanish (who are advanced

speakers of L2 English) should perform like native controls, while no or few native speakers

of Mandarin should perform in a native-like way. Thus, a closer examination of individual

performances is warranted in order to see how individual performance compares with group

results. We now turn to a discussion of how LoR relates to proficiency scores, and how both

variables relate to target-like performance.

49

Tables 4, 5, and 6 show information about each participant in each L2er group. From

left to right, the information includes: participant ID, age of arrival in the US (AoA), length

of residence (LoR), combined proficiency score, suppliance of past tense in oral responses

(Past Oral), suppliance of simple past tense in written responses (Past Written), and simple

past suppliance in the elicited description task (Elicited). Participants are listed in order of

highest proficiency score to lowest. Scores in shaded cells indicate that a given score falls

within the score range obtained by the native speaker group and thus implies native-like

performance.

Table 4. L1 Japanese group data

AoA LoR

Proficiency Past Oral Past Written

Elicited

596 26 24 98 70 92 100

777 28 21 90 100 100 100

729 27 13 92 96 100 87.5 1708 27 11 85 92 100 100

761 19 11 84 95.8 100 75

1756 21 13 80 88 100 100

205 28 1;4 78 100 100 100 494 29 5 77 95 96 87.5

798 19 9 77 96 88 100

785 19 9 76 42 77 50 891 27 1;4 76 96 80 100

As can be seen in Table 4, proficiency scores and LoR seem to be closely related; higher

(shaded) proficiency scores come from participants with over a decade of residence in the

US. This suggests that proficiency increases according to LoR. In relation to native-like

performance, a total of four participants performed in a native-like fashion on all three tasks.

It should be noted, though, that one of them (participant 205) had resided in the US for only

one year and four months and his/her proficiency score was below the controls’ range. The

table also shows that LoR and/or proficiency scores do not unequivocally predict target-like

production of simple past tense. Indeed, participant 596 has the highest LoR and proficiency

50

score (98) and yet, suppliance in the Past Oral test (70) is well below that of participant 891

(96), who has the shortest LoR and the lowest proficiency score (76). From this, we conclude

for now that LoR has an effect on proficiency, but perhaps not so much of an effect on simple

past morphology suppliance.

Table 5. L1 Spanish group data25

Participant number

Dialect AoA LoR Prof Past Oral Past Written

Elicited

747 Colombia 27 5;4 98 77 96 62.5 962 Mexico 24 17 93 96 100 100

1182 Colombia 24 4;9 93 93 100 100

293 Chile 25 8 91 96 100 100 862 Spain 32 10 91 100 100 100

265 Chile 26 1;6 89 58 100 62.5

612 Chile 28 8 88 73 100 100

782 Spain 27 14 84 100 100 100 595 Chile 32 2;4 84 70 100 55.5

1785 Colombia 36 4 82 82 96.3 75

1007 Colombia 30 4;6 79 60 85 87.5 1966 Chile 42 8;4 79 100 96 87.5

1351 Spain 25 6 77 78 100 100

In the Spanish group, table 5 shows that the relationship between LoR and proficiency score

is not entirely predictable. For example, participants 747 and participant 1351 have similar

LoRs and a large difference in proficiency scores, while other proficiency score-LoR

relationships do not seem to follow a particular pattern. However, when we look at target-like

performance across the three tasks, we do observe that participants 962, 293, 862 and 782

tend to have longer LoRs (17, 10, and 14 years, respectively).

Table 6. L1 Mandarin group data

Participant AoA LoR Prof Past Oral Past Elicited

25 One of the reviewers raises the question of the possible influence of the specific Spanish dialects.

As is well known, some Spanish dialects are more conservative than others in maintaining consonants in coda position. In the dialects represented in the study, Colombian and Mexican Spanish are particularly conservative, European Spanish (Castilian) less so, and Chilean Spanish is the least conservative. While we acknowledge that this was not controlled for and it should be in future studies, the data do not seem to suggest that those from more conservative dialects tend to keep complex codas more often than the other dialects.

51

number Written 688 25 1;4 87 92 100 100 731 22 1;6 86 89 100 66.6 247 24 1;4 85 84 89 100 464 22 3;4 82 72 100 100 869 26 6;6 81 92 96 85.7

1045 17 7 81 88 100 100 689 23 1;4 80 100 100 100

1777 23 1;6 79 65 100 87.5 986 20 3;4 77 81 96 87.5 863 22 2;6 77 60 100 45.4 865 23 1;4 76 89 92 100

1792 29 1;7 76 72 100 70 1846 23 2;6 75 73 96 85.7 1666 25 9 75 65 100 85.7 715 26 1;4 74 69 93 100

Table 6 shows the Mandarin participants’ data, and it is evident that these learners’

LoRs are much shorter than those of the Spanish and Japanese groups. We also observe that

LoR and proficiency scores do not appear to correlate. For example, the participants with the

three highest proficiency scores have a LoR of less than two years, while participants with

longer LoRs have lower proficiency scores. Additionally, participants 688 and 715 have the

same LoR but proficiency scores that differ by 13 points. Moving on to a comparison of

proficiency and performance, n examination of those participants who performed in a native-

like way in all three experimental tasks indicates that proficiency scores do not seem to

predict target-like performance. Specifically, of the four participants with target-like

performance in all three tasks (688, 1045, 689, and 865), three have a LoR of 1;4, and only

one has a LoR of seven years. Considering that Mandarin is the only group whose L1 does

not instantiate the [upast] feature, it is remarkable that 27% of the participants have attained

truly native-like performance, and that all of them performed in a native-like way on at least

one task.

52

7. Discussion and conclusion

Results from the oral picture description task are mostly in accord with the findings from the

sentence completion tasks. The general pattern observed is that native-like production of

simple past morphology is attained by around half of the participants in each L2 group while

group means are around 80%-90%. It was also found that the Mandarin lack of the [upast]

feature does not preclude its speakers from attaining target-like performance on these tasks,

indicating that these learners have access to linguistic universals in adulthood and presenting

a challenge for RDAs. Additionally, given that the performance on the written sentence

completion tasks was target-like for all L2 groups while performance on the oral production

task was not, it seems that phonological factors can and do hinder suppliance of past

morphology. Looking beyond statistical results, however, the overall trend was that the L1

Japanese participants were more accurate than the Spanish and Mandarin speakers. In the

case of Mandarin speakers, we originally thought that a lack of suppliance would be due to

syntactic reasons (i.e., retrieval of [upast]). However, we have presented evidence that even

the Spanish speakers show less-than-target performance, which cannot be attributed to lack of

the [upast] feature. The predictions of RDAs are not supported by a preponderance of all the

data taken together because: (a) Phonological issues proved to be significant in past tense

morphology suppliance for the Mandarin natives, and (b) the Mandarin and Spanish groups

patterned together in performance. Since RDAs state that problems in functional morphology

suppliance stem from a syntactic deficit that is maturationally conditioned and no specific

claims are made about the role of phonological factors, it is unclear how the hypothesis can

account for the overall tendency that functional morphology suppliance is lower in oral

production.26 It is interesting to note that phonological factors seem to be the best candidate

26 An anonymous reviewer brought up the possibility that the L1 Mandarin learners could have used

domain-general strategies to supply past tense morphology. While this is certainly a possibility (and is hard to verify), the design of both tasks was identical, as were the testing conditions. For example, when participants

53

for explaining why the L1 Spanish group performed as it did, since this group undoubtedly

has the syntactic feature for past. Conversely, the PTH is better equipped to account for such

differences in performance. Thus, we conclude that the PTH is on the right track in placing

great importance on phonological factors in oral production of functional morphology. We

believe that L1 phonological factors can strongly influence functional morphology

production, in that performance is lowered when morphology suppliance is mediated by

phonological production.

Given the evidence that points to oral suppliance affected by phonological factors, we

raise the question of whether L2 prosodic structure can be acquired, or if a learner’s

phonological grammar will fossilize rather than constructing the target PWd adjunction. The

L1 Japanese participants appear to have acquired the target structure, which indicates the

ability to build novel L2 prosodic structure via the combination of pre-existing (L1) prosodic

structures and points to at least partial access to UG in the domain of prosodic phonology.

This brings us to the question of why, 1) if all three learner groups meet one of the conditions

of Minimal Adaptation, and 2) if all three groups demonstrate acquisition of the past feature,

do several individuals in each group demonstrate acquisition of target prosodic structure, but

only the Japanese group as a whole demonstrates acquisition of these structures? This is

especially interesting given that both Spanish and Japanese disallow the word-final consonant

clusters tested in the current study. While this unexpected result is without serious

implications for the general argumentation, it is still of interest to explore different possible

explanations. In what follows, we consider language-specific possibilities, followed by

factors that could be responsible for individual variation.

completed sentences in the written task, they had the opportunity to think before completing the answer and to change their response. For the oral sentence completion task, participants were told they could take as much time as they wanted and that they could change their answers. When an answer was revised in either task, only the last response counted. Thus, we are confident that if any domain-general strategies were used, they were used similarly in both tasks.

54

One possible explanation for the Japanese group’s performance is the different L1

phonological repair mechanisms for consonant clusters employed in each language. While

Japanese repairs English borrowings with consonant clusters by epenthesizing a vowel in

between consonants, Spanish deletes one (or more) of the consonants in the cluster. Thus, we

may hypothesize that in the phonological system of the Japanese speakers, the “extra”

consonants can still be recovered in future stages of development, whereas the L1 Spanish

speakers may not have them available for future stages because they have been reduced. We

have also proposed in this study that the Japanese past tense marker is prosodified than in the

target language. However, if it is the case that the past marker in Japanese is prosodified in

the same way as English (as pointed out by one of the reviewers), then the data could more

easily be explained. Another possible explanation for inter-group variation was explored in

the previous section regarding LoR. As was seen, Japanese speakers have the longest LoRs.

However, the data do not suggest that this could have impacted past morphology suppliance,

although it did seem to impact overall proficiency.

Beyond group differences, it is of interest why each group consists of a number of

learners that tested within the native range as well as a number of those that did not. Recall

that while the Japanese speakers’ rate of suppliance was higher than that of the Mandarin and

Spanish speakers, the rates of seven of the 11 L1 Japanese participants, seven of the 13

Mandarin participants, and seven of the 15 Spanish participants were not within the native

range for the picture description task. We briefly discuss two factors that have been

hypothesized in the literature to affect L2 inflection production: 1) input factors, including

perception, salience, and frequency, and 2) performance factors, i.e., processing costs.

A link has been proposed between perception and production, whereby L2 input must be

converted to intake before production can occur (e.g., Flege 1991, 1995; Wode 1995, 1997).

Researchers such as Brown (1998) and Kabak & Idsardi (2007) claim that such conversion

55

can be blocked by the L1 phonology, thus impeding intake and therefore production. In fact,

there is ample evidence that L1 phonotactic constraints can impede accurate L2 perception

(e.g., Dupoux, Kakehi, Hirose, Pallier & Meller, 1999; Kabak & Idsardi 2007; Dupoux,

Parloto, Frota, Hirose & Peperkamp 2011). That being said, if it is the case that perception

indeed precedes production, past inflection production data from a number of studies (e.g.,

Hawkins & Liszka 2003; Goad & White 2006, Lardiere 2007; Bonner 2013) would indicate

that learners are indeed able to perceive coda consonant clusters that violate L1 phonotactic

constraints. Indeed, recent research (e.g., Martahardjono, Valian & Klein 2011; Bonner &

Martahardjono 2012) shows that L1 constraints do not affect perception, but rather that

salience and processing costs are the culprits when it comes to inflectional variability.

Given the nature of the tokens that were used in the tasks, it is conceivable that salience is

partially responsible for non-native-like performance. First, the targets employed were

utterance medial or utterance final, which Barcroft & Van Patten (1997) found to be less

salient than an utterance initial target (although we recognize that the stop release in word-

final position should aid perception, see Goad 2011, for discussion). Second, targets

contained the non-syllabic voiceless allomorph [t], also found to be less salient than syllabic

and voiced [Id] (Bonner & Martahardjono 2012). Third, some CC targets contained adjacent

segments with identical manners of articulation, which has been shown to affect

perceptibility (e.g., Adams 2004). Finally, Martahardjono, Valian & Klein (2012) found that

bound morphemes are less salient than free morphemes. Thus, it is conceivable that some

learners have not converged on the target because the input is not salient enough in certain

contexts (see Archibald 2009 for a discussion of the importance of salience in overcoming

L1/L2 differences).

A growing body of research (e.g., Trenkic 2007; Hopp 2009; Martohardjono, Valian

& Klein 2012; Bonner 2013) proposes that variability, even at advanced levels of proficiency,

56

is a result of processing load effects. Hopp (2009) found that in tasks with a higher

processing demand, near-native L2ers evidenced processing deficiencies similar to those of

native speakers that were tested. Bonner’s (2013) findings echo those of Hopp: he found a

significant effect of sentence length in both perception and production of past tense

inflection. Considering the findings reported in this section, the issues of L1 perceptual

constraints, salience, and processing load all warrant further exploration and is currently

underway (Authors, in preparation).

To conclude, we believe that the data clearly show that uninterpretable features are

retrievable from the universal inventory in adulthood, but in order for these features to

manifest in a fully native-like way on the surface, additional knowledge beyond syntactic

representation is necessary. Beyond the fact that Mandarin and Spanish speakers seem to

parallel each others’ performance, further compelling evidence to be highlighted comes from

the fact that 4 out of the 15 Mandarin speakers supplied past tense morphology in a target-

like fashion in all tasks, which casts doubts on the (absolute) unavailability of new

uninterpretable features as well as the impossibility of constructing L2 prosodic structure. In

the end, perhaps not all uninterpretable features are equal, in the sense that they interact

differently with phonological properties.

57

References

Adams, Kent. 2004. Variability in the perception of past tense -ed by L2 learners of English.

Paper presented at SUNY/CUNY/NYU 5, May 1, Stony Brook, SUNY. Adger, David. 2003. Core syntax. Oxford: Oxford University Press.

Alers-Valentin, Hilton. 2000. The prosodic structure of the Spanish verb: Evolution and configuration. Amherst, MA: University of Massachusetts dissertation.

Archibald, John. 2009. Phonological feature re-assembly and the importance of phonetic

cues. Second Language Research 25(2). 231-233. Barcroft, Joe & Bill VanPatten. (1997). Acoustic salience of grammatical forms: tThe effect

of location, stress, and boundedness on Spanish L2 input processing. In William Glass and Ana Teresa Perez-Leroux (eds.), Contemporary perspectives on the acquisition of Spanish. Vol. 2: Production, processing, and comprehension, 109-121. Somerville, MA: Cascadilla Press.

Boersma, Paul & David Weenink. 2014. Praat: doing phonetics by computer [Computer

program]. http://www.praat.org/ (18 May, 2014) Bonner, Timothy. 2013. Systematic asymmetries in perception and production of L2

inflections in Mandarin L2 learners of English: The effects of phonotactics, salience, and processing pressure on inflectional variability. New York, NY: The Graduate Center, CUNY dissertation.

Bonner, Timothy & Gita Martohardjono. 2012. Performance factors trump representational

deficits: Perception and production of English inflections by L1 Mandarin speakers. In Alia Biller, Esther Chung and Amelia Kimball (eds.), Proceedings of the 36th annual Boston University Conference on Language Development [BUCLD 36], 74-86. Somerville, MA: Cascadilla Press.

Brown, Cynthia 1998. The role of the L1 grammar in the L2 acquisition of segmental

structure. Second Language Research 14(2). 136-193. Bybee, Joan. 1995. Regular morphology and the lexicon. Language and Cognitive processes

10. 425-55. Clahsen, Harald. (1999). Lexical entries and rules of language: a multidisciplinary study of

German inflection. Behavioral and Brain Sciences 22. 991-1060. Clahsen, Harald & Monika Rothweiler. (1993). Inflectional rules in children’s grammars:

evidence from the development of participles in German. In Geert Booij and Joop van Marle (eds.), Morphology Yearbook 1992, 1-34. Houten, Netherlands: Springer.

Colina, Sonia. 2006. Optimality-theoretic advances in our understanding of Spanish syllable

structure. In Fernando Martinez-Gil and Sonia Colina (eds.), Optimality-theoretic studies in Spanish phonology, 172-204. Amsterdam: John Benjamins.

58

Davies, Mark. 2008. Corpus of contemporary American English.

http://corpus.byu.edu/coca/ (21 May, 2014) Dupoux, Emmanuel, Kazuhiko Kakehi, Yuki Hirose, Christophe Pallier & Jacques Mehler.

1999. Epenthetic vowels in Japanese: A perceptual illusion? Journal of Experimental Psychology- Human Perception and Performance 25(6). 1568-1578.

Dupoux, Emmanuel, Erika Parlato, Sonia Frota, Yuki Hirose & Sharon Peperkamp. 2011.

Where do illusory vowels come from? Journal of Memory and Language 64(3). 199-210.

Duanmu, Sam. 2007. The phonology of standard Chinese. Oxford: Oxford University Press. Flege, James Emil 1991. The interlingual identification of Spanish and English vowels:

orthographic evidence. Quarterly Journal of Experimental Psychology 43A. 701-731. Flege, James Emil. 1995. Second language speech learning: Theory, findings, and problems.

In Winifred Strange (ed.), Speech perception and linguistic experience: Issues in cross-language research, 233-277. Timonium, MD: York Press.

Franceschina, Florencia. 2002. Case and (phi)-feature agreement in advanced L2 Spanish

grammars. EUROSLA Yearbook 2. 71-86. Franceschina, Florencia. 2005. Fossilized second language grammars. Amsterdam: John

Benjamins. Goad, Heather. 2011. The L2 acquisition of functional morphology: Why syntacticians need

phonologists. In Julia Herschensohn & Darren Tanner (eds.), Proceedings of the 11th Generative Approaches to Second Language Acquisition Conference [GASLA 2011], 1-16. Somerville, MA: Cascadilla Press.

Goad, Heather, Lydia White & Jeffrey Steele. 2003. Missing inflection in L2 acquisition:

Defective syntax or L1-constrained prosodic representations? The Canadian Journal of Linguistics/La Revue Canadienne De Linguistique 48(3-4). 243-263.

Goad, Heather, Lydia White & Joyce Bruhn de Garavito. 2011. Prosodic transfer at different

levels of structure: The L2 acquisition of Spanish plurals. In Nick Danis, Kate Mesh & Hyunsuk Sung (eds.), Online Proceedings of the 35th Annual Boston University Conference on Language Development [BUCLD 35]. http://www.bu.edu/bucld/files/2011/05/35-Goad-White-Bruhn-de-Garavito.pdf (21 May, 2014)

Goad, Heather & Lydia White. 2006. Ultimate attainment in interlanguage grammar: A

prosodic approach. Second Language Research 22. 243-268. Harris, James. 1983. Syllable structure and stress in Spanish: A nonlinear analysis. Cambridge,

MA: MIT Press.

59

Harris, James. 1987. The accentual patterns of verb paradigms in Spanish. Natural Language and Linguistic Theory 5. 61-90.

Hawkins, Roger & Hajime Hattori. 2006. Interpretation of English multiple wh-questions by

Japanese speakers: a missing uninterpretable feature account. Second Language Research 22(3). 269-301.

Hawkins, Roger & Sarah Liszka. 2003. Locating the source of defective past tense markingin advanced L2 English speakers. In Roeland van Hout, Aafke Hulk, Folkert Kuiken & Richard Towell (eds.), The lexicon-syntax interface in second language acquisition, 21-44. Amsterdam; Netherlands: John Benjamins.

Hopp, Holger. 2009. The syntax-discourse interface in near-native L2 acquisition: Off-line

and on-line performance. Bilingualism: Language and Cognition 12. 463-483. Hualde, José Ignacio, Antxon Olarrea & Ana María Escobar. 2003. Introducción a la

lingüística hispánica. Cambridge, UK: Cambridge University Press. Ito, Junko & Armin Mester. 2003. Weak layering and word binarity. In Takeru Honma,

Masao Okazaki, Toshiyuki Tabata and Shin-ichi Tanaka (eds.), A new century of phonology and phonological theory. A festschrift for Professor Shosuke Haraguchi on the occasion of his sixtieth birthday. 26-65. Tokyo: Kaitakusha.

Ito, Junko & Armin Mester. 2007. Prosodic adjunction in Japanese compounds. MIT Working

Papers in Linguistics 55: Formal approaches to Japanese linguistics 4. 97-111. Kabak, Baris & William Idsardi. 2007. Perceptual distortions in the adaptation of English

consonant clusters: syllable structure or consonantal contact constraints? Language and Speech 50(1). 23-52.

Kuroda, Sige-Yuki. 1979. On Japanese passives. In George Bedell, Eichi Kobayashi and

Masatake Muraki (eds.), Explorations in linguistics: Papers in honor of Kazuko Inoue, 305-347. Tokyo: Kaitakusha.

Lardiere, Donna. 1998a. Case and tense in the ‘fossilized’ steady state. Second Language

Research 14. 1-26. Lardiere, Donna. (1998b). Dissociating syntax from morphology in a divergent L2 end-state

grammar. Second Language Research 14(4). 359-375. Lardiere, Donna. 2000. Mapping features to form in second language acquisition. In John

Archibald (ed.), Second language acquisition and linguistic theory, 102-29. Oxford: Blackwell.

Lardiere, Donna. 2007. Ultimate attainment in second language acquisition: a case study.

Mahwah, NJ: Lawrence Erlbaum Associates. Lardiere, Donna. 2009. Some thoughts on a contrastive analysis of features in second language

acquisition. Second Language Research 25(2). 171–225.

60

Lobeck, Anne. 2000. Discovering grammar: an introduction to English sentence structure. New York: Oxford University Press.

Marshall, Chloë & Heather van der Lely. 2006. A challenge to current models of past tense

inflection: the impact of phonotactics. Cognition 100(2). 302-320. Martohardjono, Gita, Virginia Valian & Elaine Klein. 2012. The tense puzzle in second

language acquisition: What part representation? What part performance? Ms. The Graduate Center, CUNY.

McCarthy, Corrine. 2008. Morphological variability in the comprehension of agreement: an

argument for representation over computation. Second Language Research 24(4). 459–486.

McClelland, James & Karalyn Patterson. 2002. 'Words or rules' cannot exploit the regularity

in exceptions: Reply to Pinker and Ullman. Trends in Cognitive Sciences 6(11). 464-465.

McCawley, James. 1968. The Phonological component of a grammar in Japanese. The

Hague: Mouton. Nation, Paul. 1990. Teaching and learning vocabulary. Boston: Heinle and Heinle. Núñez Cedeño, Rafael & Alfonso Morales-Front. 1999. Fonología generativa

contemporánea de la lengua española. Washington DC: Georgetwon University Press.

Okuwaki, Natsumi. 2000. Japanese –ta as an auxiliary verb. Ms. University of Essex. Ota, Mitsuhiko. 1999. Phonological theory and the acquisition of prosodic structure: Evidence

from child Japanese. Dissertation Abstracts International, A: The Humanities and Social Sciences 6(12). 4752-A-4753-A.

Peperkamp, Sharon. 1997. Prosodic words. The Hague: Holland Academic Graphics. Pinker, Steven. 1991. Rules of Language. Science 253. 530-35. Pinker, Steven & Michael Ullman. 2002. The past and future of the past tense. Trends in

Cognitive Sciences 6(11), 456-463. Piñeros, Carlos Eduardo. 2008. Estructura de los sonidos del español. Upper Saddle River,

N.J.: Pearson Prentice Hall. Prasada, Sandeep & Steven Pinker. 1993. Generalization of regular and irregular

morphological patterns. Language and Cognitive Processes 8. 1-56. Prevost, Phillipe & Lydia White. 2000. Missing surface inflection or impairment in second

language acquisition? evidence from tense and agreement. Second Language Research 16(2).103-133.

61

Rothman, Jason. 2007. Sometimes they use it, sometimes they don’t: An epistemological discussion of L2 Morphological production and its use as a competence measurement. Applied Linguistics 28(4). 609-614.

Schwartz, Bonnie. 1998. The second language instinct. Lingua 106. 133-160. Schwartz, Bonnie & Rex Sprouse. 1996. L2 cognitive state and the Full Transfer/Full Access

Model. Second Language Research 12. 40-72. Slabakova, Roumyana. 2008. Meaning in the second language. Berlin/New York: Mouton de

Gruyter. Soga, M. (1983). Tense and aspect in modern colloquial Japanese. Vancouver: University of

British Columbia Press. Solt, Stephanie, Yana Pugach, Elaine C. Klein, Kent Adams, Iglika Stoyneshka & Tamara

Rose. 2004. L2 perception and production of the English regular past: evidence of phonological effects. In Alejna Brugos, Linnea Micciulla & Christine Smith (eds.), Proceedings of the 36th annual Boston University Conference on Language Development [BUCLD 28], 553-564. Somerville, MA: Cascadilla Press.

Selkirk, Elizabeth. 1997. The prosodic structure of function words. In James Morgan &

Katherine Demuth (eds.), Signal to syntax: Bootstrapping from speech to grammar in early acquisition, 187-213. Mahwah, NJ: Lawrence Erlbaum.

Tsimpli, Ianthi & Maria Dimitrakopoulou. 2007. The Interpretability Hypothesis: Evidence

from wh-interrogatives in second language acquisition. Second Language Research 23(2). 215-242.

Tsujimura, N. 2007. An introduction to Japanese linguistics, 2nd edn. Oxford: Blackwell. Vance, Timothy. 1987. An Introduction to Japanese phonology. State University of New York

Press, Albany. Vance, Timothy. 2008. The sounds of Japanese. Cambridge: Cambridge University Press. Wode, Henning. (1995). Speech perception, language acquisition, and linguistics: some

mutual implications. In Winifred Strange (ed.), Speech perception and linguistic experience: Issues in cross-language research, 321-347. Timonium, MD: York Press.

Wode, Henning. (1997). Perception and production in learning to talk. In S.J. Hannahs and

Martha Young-Scholten (eds.), Focus on phonological acquisition, 153-191. Amsterdam: John Benjamins.

62

Appendix 1: Picture description Jane’s day off

Yesterday it was Sunday so Jane took the day off. Please tell the story of her day

off. Start your story with: Yesterday, it was Jane’s day off…

Morning

sleep – 10:00 work out -10:30 buy groceries - 11:15 Afternoon

wash - 1:00 clean – 1:30 bake - 2:00

Evening

walk 5:00 drink - 6:30 go to the movies – 9:00

63

Appendix 2: sentence completion tests.

[ld] coda type 1. Sealed She wrote her comments and suggestions, placed them in an envelope and then _______ it (20 syl)

[seal kneel work] Mary put everything back in her room quickly, closed the door and _______ it with tape. (20 syl)

[call send seal] 2. filled Johnny had a terrible headache, so he _______ a glass with water and took two aspirins. (21 syl) [fill write type] In 1950, all computers in existence were huge. Each one _______ a whole room! (21 syl)

[send fill smell] 3. Killed Surprisingly, last year human diseases _______ an exceptional number of lions. (22 syl)

[lose yell kill] Last summer in Colorado, the West Nile virus _______ more than 100 thousand birds (20 syl)

[call crash kill] 4. Yelled Since they were losing the game, coach Jason was mad and _______ angrily at his team. (20 syl)

[mail write yell] Dylan was desperately looking for his wife, so he opened the window and _______ “Anne!!” (22 syl)

[run sneeze yell] 5. Rolled Well, his car slipped on the wet road, and then it _______ over and crashed into a ditch (20 syl)

[roll drive feel] John is helpful but never listens. I asked him to clean the carpet, but he _______ it up instead. (24 syl)

[drive feed roll] Monomorphemes [ld] 1. build

64

The city also plans to upgrade the aquarium and _______ a new park next year. (20 syl)

[build call field] The property is owned by developers planning to _______ a casino resort (21 syl)

[touch build hear] 2. wild One-fourth of seafood harvested from the _______ is not used for human consumption (20 syl)

[wild book sleep ] She is proud of herself for doing something that was on the _______ and crazy side. (20 syl) [book wild shirt] 3. field If student athletes don’t do well in their tests, they can’t go to their basketball court or football _______.

[field lab work] (24 syl) Gender equality has long been an issue in the _______ of education (20 syl)

[man cup field] 4. mild Go to a hospital if you have severe symptoms, but stay home if your symptoms are _______ (22 syl)

[nice than mild] Many economists thought that the impact from the US crisis would be globally _______ (23 syl)

[car mild fresh] 5. fold Usually when I'm done painting, I _______ a paper towel and wipe off the excess paint. (22 syl)

[fold call eat] First, slice one onion and then _______ a sheet of foil in half to put the slices on it. (21 syl)

[buy fold smell] [nd] coda type 1. joined I wanted to get closer to her, so I even ________ a sculpture club to learn her craft (22 syl)

[eat see join] We invited faculty to be a part of the spiritual life group and everyone ________. (22 syl)

[throw see join]

65

2. cleaned John was in charge of building administration. He checked the locks nightly and ________ everything.

[clean train cut] (23 syl) When the factory found out about the water pollution, they ________ up the river. (20 syl)

[kill clean throw] 3. trained As many European basketball players, Ivan left his country and ________ in the USA. (23 syl)

[train buy catch] “John said that to prepare for his fight last September, he ________ everyday for three months”. (20 syl)

[fall train have] 4. ruined I had a beautiful restaurant but yesterday the police ________ it while chasing a suspect. (24 syl)

[buy eat ruin] The Kansas team had a very bad coach, who actually ________ a whole season of games. (21 syl) [write ruin call] 5. gained Last month, I stopped eating healthy food; I started eating doughnuts and ________ about ten pounds. (21 syl) [gain drink train] Pakistan was created when India ________ its independence in 1947. (22 syl)

[house gain call] Monomorphemes [nd] (2 1. pound Few people know that a distance runner needs more protein per ________ of body weight than a weight lifter. (24 syl) [pound fork work] He was taken into custody after a search of his car showed half a ________ of drugs (22 syl) [book pound field] 2. send “Well, everything is by phone, and I have no idea where to ________ a complaint letter” (22 syl) [buy fry send]

66

If you want a paper version of this article, please ________ a self-addressed, stamped envelope. (22 syl). [blend catch send] 3. blend If you want a good sweater, only buy one made of wool or cashmere, or a ________ of both. (22 syl).

[team pound blend] What makes for a good dessert? I believe, a ________ of anything chocolate and peanut butter (23 syl) [blend leaf cold] 4. pond My dad and I were driving, and suddenly he stopped the car and pointed at a nice frozen ________ (24 syl) [pond sky truck] They installed a waterfall and also, they used pretty flowers to cover the ________ (21 syl) [fall pond cry] 5. blind My sister always wanted to become a child psychologist for the deaf and ________. (21 syl)

[soon blind home] A crisis exists in the US because of the shortage of teachers for students who are ________. (24 syl)

[more draft blind] Past forms [st] 1. passed Sacred lands have been protected for 10 years now thanks to the law that the Supreme Court ________. (22 syl) [buy pass say] The regulations in his basketball team are strict, and he ________ a drug test just yesterday. (23 syl) [pass sleep type] 2. pressed Paul refused to attend the meeting and ________ other people to stay away, too. (20 syl) [above press know] Jim was in the hospital, and whenever he felt worse, he just ________ a button for help. (22 syl) [rent throw press] 3. kissed This very young student said that the first time she ________ a guy, she thought it was disgusting. (21 syl) [kiss write call]

67

I was driving through the town with my father and he showed me where he first ________ a girl. (21 syl)

[fall kiss fly] 4. blessed She hugged everyone as if preparing for a long journey, and she ________ us over and over (24 syl). [buy bless close] The nun prayed for all of us and ________ all of us, especially the sickly ones. (20 syl)

[bless see pass] 5. missed My dad was a wonderful father, he never ________ a soccer game or a hockey game. (22 syl). [eat write miss] I ended up in the emergency room with salmonella, and ________ a whole week of work. (24 syl) [see clap miss] Monomorphemes [st] 1. test She said that only one in three fourth graders here can pass the state's basic math ________ (20 syl) [room test card] Based on the research, the student or team should plan, build, and ________ a working model of the system (24 syl) [fine cry test] 2. list They told me to drink plenty of water and also gave me a ________ of healthful foods to eat. (24 syl) [list dirt help] I never liked my high school teacher. Once, she gave us a ________ of a hundred words to memorize. (24 syl) [rug sea list] 3. dust Don’t apply solvent, just use a cloth moistened with water to clean off the ________ after sanding (23 syl) [hole dust air] Most big cities are incredibly polluted because factories contribute to smog and ________. (24 syl) [car peace dust] 4. beast He is big, and his feet are huge. He is like a creature from mythology, like half man, half ________. (24 syl) [house beast book] The story makes no mention of the hero, though it suggests that someone killed the ________. (21 syl) [rock paint beast]

68

5. nest The old lady found several tiny slips of paper in a bird ________ in her backyard. (21 syl) [rust tribe nest] Her heart was filled with joy. She loved looking at the baby bird in the artfully hidden ________ (23 syl) [nest old frame] SIMPLE CODA TOKENS IN PAST TENSE 1. Played On our soccer team, we like challenges, so last weekend we ___________ against the best team in the city. (25 syl) [play want try] Yesterday, we cleaned the playground, and today in the morning, my kids ___________ in the sandbox for hours. (25 syl) [cook sell play] 2. Fried My mom cooked all day long. For lunch, she made a salad, and then she ___________ a chicken. (20 syl) [shift fry run] I was very hungry this morning but didn’t have much time, so I just ___________ an egg and ate it. (23 syl) [fry know cut] 3. Tried At the picnic there were 5 types of pie but I couldn’t choose just one, so I ___________ a piece of each. (23 syl) [scan try write] After looking at the options at their health club, Camille and Kent ___________ a fitness yoga class. (23 syl) [make try cut] 4. Paid Last weekend, Allan was caught by the police driving while drunk. He ___________ a $1,000 fine. (23 syl) [write call pay] John always cheats. Yesterday, for his math exam, he ___________ a friend to take the exam for him. (23 syl) [pull pay lift] 5. cried Mary was talking on the phone and was told that her father had died. She got off the phone and ___________. (24 syl) [cry write roll] I'm not emotional at all, but I admit that when my brother got married, I ___________. (22 syl) [mark cry run]

69

6. stayed Stan flew back from vacation in Mexico on Sunday, and he ___________ at Rob's Sunday night. (20 syl) [drive stay cry] Because of the snowstorm last week, schools were closed but only some of the children ___________ at home. (22 syl) [stay buy go]

70

Appendix 3. Monomorphemes and simple passt forms (numbers indicate natural logarithm)

[ld] [nd] [st] Past form Monomorpheme Past form Monomorpheme Past form Monomorpheme

filled (10.28)

build (10.52) joined (10.15)

send (10.42) passed (10.65)

test (11)

killed (10.75)

field (11.2) trained (9.6) blind (9.48) missed (9.74)

list (10.78)

rolled (9.43)

wild (10.21) gained (9.3) pound (9.03) pressed (9.3)

dust (9.7)

yelled (8.6) mild (8.78) cleaned (8.41)

blend (8.7) kissed (8.72)

priest (9.14)

sealed (8.4) bald (8.14) ruined (8.25) pond (8.7) blessed (8.47)

nest (8.56)

p = 0.6 p = 0.74 p = 0.36